diff --git a/.github/workflows/codegen-drift.yml b/.github/workflows/codegen-drift.yml new file mode 100644 index 00000000..a26caf5e --- /dev/null +++ b/.github/workflows/codegen-drift.yml @@ -0,0 +1,61 @@ +name: Codegen Drift Check + +on: + pull_request: + types: [opened, synchronize, reopened] + paths: + - 'api/**' + - 'pkg/apis/**' + - 'pkg/generated/**' + - 'internal/crdinstall/manifests/**' + - 'packages/system/cozystack-controller/definitions/**' + - 'packages/system/application-definition-crd/definition/**' + - 'packages/system/backup-controller/definitions/**' + - 'packages/system/backupstrategy-controller/definitions/**' + - 'hack/update-codegen.sh' + - 'hack/boilerplate.go.txt' + - 'Makefile' + - 'go.mod' + - 'go.sum' + - '.github/workflows/codegen-drift.yml' + +concurrency: + group: codegen-drift-${{ github.workflow }}-${{ github.event.pull_request.number }} + cancel-in-progress: true + +jobs: + codegen-drift: + name: Verify generated code is up to date + runs-on: ubuntu-22.04 + steps: + - name: Checkout code + uses: actions/checkout@v4 + + - name: Set up Go + uses: actions/setup-go@v5 + with: + go-version-file: go.mod + cache: true + + - name: Pre-fetch k8s.io/code-generator module + # hack/update-codegen.sh sources kube_codegen.sh from the Go module cache. + # The module is not declared in go.mod, so fetch it explicitly at the + # version pinned in the script. + run: | + version=$(grep -oP 'code-generator@\Kv[0-9.]+' hack/update-codegen.sh) + tmpdir=$(mktemp -d) + cd "$tmpdir" + go mod init codegen-fetch + go get "k8s.io/code-generator@${version}" + + - name: Run make generate + run: make generate + + - name: Fail on drift + run: | + if [ -n "$(git status --porcelain)" ]; then + echo "::error::'make generate' produced changes. Run 'make generate' locally and commit the result." + git status --short + git diff --color=always + exit 1 + fi diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml index ac0f7e30..6492b92d 100644 --- a/.pre-commit-config.yaml +++ b/.pre-commit-config.yaml @@ -1,6 +1,17 @@ repos: - repo: local hooks: + - id: run-make-generate-root + name: Run 'make generate' at repo root + entry: | + flock -x .git/pre-commit.lock sh -c ' + echo "Running make generate at repo root" + make generate || exit $? + git diff --color=always | cat + ' + language: system + files: ^(api/|pkg/apis/|pkg/generated/|internal/crdinstall/manifests/|packages/system/cozystack-controller/definitions/|packages/system/application-definition-crd/definition/|packages/system/backup-controller/definitions/|packages/system/backupstrategy-controller/definitions/|hack/update-codegen\.sh$|hack/boilerplate\.go\.txt$|Makefile$) + pass_filenames: false - id: run-make-generate name: Run 'make generate' in all app directories entry: | diff --git a/Makefile b/Makefile index 10a55314..acc9b6f6 100644 --- a/Makefile +++ b/Makefile @@ -22,6 +22,7 @@ build: build-deps make -C packages/system/lineage-controller-webhook image make -C packages/system/cilium image make -C packages/system/linstor image + make -C packages/system/linstor-gui image make -C packages/system/kubeovn-webhook image make -C packages/system/kubeovn-plunger image make -C packages/system/dashboard image diff --git a/api/apps/v1alpha1/kubernetes/types.go b/api/apps/v1alpha1/kubernetes/types.go index ea4a2b64..88a3c97e 100644 --- a/api/apps/v1alpha1/kubernetes/types.go +++ b/api/apps/v1alpha1/kubernetes/types.go @@ -69,6 +69,9 @@ type Addons struct { // NVIDIA GPU Operator. // +kubebuilder:default:={} GpuOperator GPUOperatorAddon `json:"gpuOperator"` + // HAMi GPU virtualization middleware. + // +kubebuilder:default:={} + Hami HAMiAddon `json:"hami"` // Ingress-NGINX controller. // +kubebuilder:default:={} IngressNginx IngressNginxAddon `json:"ingressNginx"` @@ -160,6 +163,15 @@ type GatewayAPIAddon struct { Enabled bool `json:"enabled"` } +type HAMiAddon struct { + // Enable HAMi (requires GPU Operator). + // +kubebuilder:default:=false + Enabled bool `json:"enabled"` + // Custom Helm values overrides. + // +kubebuilder:default:={} + ValuesOverride k8sRuntime.RawExtension `json:"valuesOverride"` +} + type Images struct { // Image used by the wait-for-kubeconfig init container. Empty falls back to images/busybox.tag. // +kubebuilder:default:="" diff --git a/api/apps/v1alpha1/kubernetes/zz_generated.deepcopy.go b/api/apps/v1alpha1/kubernetes/zz_generated.deepcopy.go index 2cbe3ba1..e021fbaa 100644 --- a/api/apps/v1alpha1/kubernetes/zz_generated.deepcopy.go +++ b/api/apps/v1alpha1/kubernetes/zz_generated.deepcopy.go @@ -49,6 +49,7 @@ func (in *Addons) DeepCopyInto(out *Addons) { in.Fluxcd.DeepCopyInto(&out.Fluxcd) out.GatewayAPI = in.GatewayAPI in.GpuOperator.DeepCopyInto(&out.GpuOperator) + in.Hami.DeepCopyInto(&out.Hami) in.IngressNginx.DeepCopyInto(&out.IngressNginx) in.MonitoringAgents.DeepCopyInto(&out.MonitoringAgents) in.Velero.DeepCopyInto(&out.Velero) @@ -135,6 +136,7 @@ func (in *ConfigSpec) DeepCopyInto(out *ConfigSpec) { } in.Addons.DeepCopyInto(&out.Addons) in.ControlPlane.DeepCopyInto(&out.ControlPlane) + out.Images = in.Images } // DeepCopy is an autogenerated deepcopy function, copying the receiver, creating a new ConfigSpec. @@ -260,6 +262,37 @@ func (in *GatewayAPIAddon) DeepCopy() *GatewayAPIAddon { return out } +// DeepCopyInto is an autogenerated deepcopy function, copying the receiver, writing into out. in must be non-nil. +func (in *HAMiAddon) DeepCopyInto(out *HAMiAddon) { + *out = *in + in.ValuesOverride.DeepCopyInto(&out.ValuesOverride) +} + +// DeepCopy is an autogenerated deepcopy function, copying the receiver, creating a new HAMiAddon. +func (in *HAMiAddon) DeepCopy() *HAMiAddon { + if in == nil { + return nil + } + out := new(HAMiAddon) + in.DeepCopyInto(out) + return out +} + +// DeepCopyInto is an autogenerated deepcopy function, copying the receiver, writing into out. in must be non-nil. +func (in *Images) DeepCopyInto(out *Images) { + *out = *in +} + +// DeepCopy is an autogenerated deepcopy function, copying the receiver, creating a new Images. +func (in *Images) DeepCopy() *Images { + if in == nil { + return nil + } + out := new(Images) + in.DeepCopyInto(out) + return out +} + // DeepCopyInto is an autogenerated deepcopy function, copying the receiver, writing into out. in must be non-nil. func (in *IngressNginxAddon) DeepCopyInto(out *IngressNginxAddon) { *out = *in diff --git a/api/apps/v1alpha1/postgresql/types.go b/api/apps/v1alpha1/postgresql/types.go index fa85d51f..a2ff77a8 100644 --- a/api/apps/v1alpha1/postgresql/types.go +++ b/api/apps/v1alpha1/postgresql/types.go @@ -92,6 +92,9 @@ type Bootstrap struct { // Timestamp (RFC3339) for point-in-time recovery; empty means latest. // +kubebuilder:default:="" RecoveryTime string `json:"recoveryTime,omitempty"` + // Barman server name (S3 path prefix) used by the original cluster when writing backups. Set this only when the original cluster had an explicit barmanObjectStore.serverName that differed from its Kubernetes resource name. + // +kubebuilder:default:="" + ServerName string `json:"serverName,omitempty"` } type Database struct { diff --git a/api/apps/v1alpha1/vminstance/types.go b/api/apps/v1alpha1/vminstance/types.go index 2bb059c6..9c48724a 100644 --- a/api/apps/v1alpha1/vminstance/types.go +++ b/api/apps/v1alpha1/vminstance/types.go @@ -26,6 +26,9 @@ type ConfigSpec struct { // Ports to forward from outside the cluster. // +kubebuilder:default:={22} ExternalPorts []int `json:"externalPorts,omitempty"` + // Whether to accept ICMP traffic to the VM in PortList mode (preserves ping and PMTU discovery). No effect in WholeIP mode. Default true so ping behaves as users expect even when port filtering is in effect. + // +kubebuilder:default:=true + ExternalAllowICMP bool `json:"externalAllowICMP"` // Requested running state of the VirtualMachineInstance // +kubebuilder:default:="Always" RunStrategy RunStrategy `json:"runStrategy"` diff --git a/api/backups/v1alpha1/zz_generated.deepcopy.go b/api/backups/v1alpha1/zz_generated.deepcopy.go index 89f6171f..61b8f839 100644 --- a/api/backups/v1alpha1/zz_generated.deepcopy.go +++ b/api/backups/v1alpha1/zz_generated.deepcopy.go @@ -620,6 +620,11 @@ func (in *RestoreJobSpec) DeepCopyInto(out *RestoreJobSpec) { *out = new(v1.TypedLocalObjectReference) (*in).DeepCopyInto(*out) } + if in.Options != nil { + in, out := &in.Options, &out.Options + *out = new(runtime.RawExtension) + (*in).DeepCopyInto(*out) + } } // DeepCopy is an autogenerated deepcopy function, copying the receiver, creating a new RestoreJobSpec. diff --git a/dashboards/gpu/gpu-efficiency.json b/dashboards/gpu/gpu-efficiency.json new file mode 100644 index 00000000..28a24568 --- /dev/null +++ b/dashboards/gpu/gpu-efficiency.json @@ -0,0 +1,819 @@ +{ + "uid": "gpu-efficiency", + "title": "GPU Efficiency Score", + "description": "Tensor saturation, util/watt and throttling — reveals inefficient GPU workloads", + "tags": [ + "gpu", + "efficiency", + "finops" + ], + "timezone": "browser", + "editable": true, + "graphTooltip": 1, + "time": { + "from": "now-1h", + "to": "now" + }, + "fiscalYearStartMonth": 0, + "schemaVersion": 42, + "panels": [ + { + "type": "row", + "collapsed": false, + "title": "Overall efficiency metrics", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 0 + }, + "id": 1, + "panels": [] + }, + { + "type": "stat", + "id": 2, + "targets": [ + { + "expr": "avg(gpu:tensor_saturation:avg5m) * 100", + "refId": "A" + } + ], + "title": "Avg Tensor Saturation", + "description": "Mean tensor core saturation across all GPUs. \u003c10% means GPUs are used inefficiently (workloads could move to CPU or optimize their code). Cluster-wide — DCGM metrics cannot be attributed to workload namespaces.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 5, + "w": 8, + "x": 0, + "y": 1 + }, + "repeatDirection": "h", + "options": { + "graphMode": "area", + "colorMode": "background", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "percent", + "min": 0, + "max": 100, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "red" + }, + { + "value": 10, + "color": "orange" + }, + { + "value": 30, + "color": "yellow" + }, + { + "value": 60, + "color": "green" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "stat", + "id": 3, + "targets": [ + { + "expr": "avg(gpu:util_per_watt:avg5m)", + "refId": "A" + } + ], + "title": "Avg Utilization per Watt", + "description": "NVML utilization % per watt across all GPUs. Higher value = more efficient workload. Cluster-wide — DCGM metrics cannot be attributed to workload namespaces.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 5, + "w": 8, + "x": 8, + "y": 1 + }, + "repeatDirection": "h", + "options": { + "graphMode": "area", + "colorMode": "background", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "none", + "decimals": 3, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "red" + }, + { + "value": 0.5, + "color": "yellow" + }, + { + "value": 1, + "color": "green" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "stat", + "id": 4, + "targets": [ + { + "expr": "avg(gpu:power_throttle_fraction:rate5m)", + "refId": "A" + } + ], + "title": "Avg Power Throttling", + "description": "Fraction of time GPUs hit the TDP cap and lose performance. \u003e5% means tenants underutilize billed FLOPS. Cluster-wide — rule aggregates by (Hostname, gpu, UUID) and drops namespace label.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 5, + "w": 8, + "x": 16, + "y": 1 + }, + "repeatDirection": "h", + "options": { + "graphMode": "area", + "colorMode": "background", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "percentunit", + "decimals": 2, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "green" + }, + { + "value": 0.05, + "color": "yellow" + }, + { + "value": 0.2, + "color": "red" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "row", + "collapsed": false, + "title": "NVML vs Tensor (mismatch detector)", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 6 + }, + "id": 10, + "panels": [] + }, + { + "type": "timeseries", + "id": 11, + "targets": [ + { + "expr": "DCGM_FI_DEV_GPU_UTIL", + "legendFormat": "{{Hostname}}/{{gpu}}", + "refId": "A" + } + ], + "title": "NVML GPU Utilization", + "description": "Classic utilization metric. Shows activity of any engine (SM, copy, encoder). Cluster-wide — DCGM namespace is the exporter's own namespace, not the workload namespace.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 0, + "y": 7 + }, + "repeatDirection": "h", + "options": { + "legend": { + "displayMode": "table", + "placement": "bottom", + "showLegend": false, + "calcs": [ + "mean", + "max" + ] + }, + "tooltip": { + "mode": "multi", + "sort": "" + } + }, + "fieldConfig": { + "defaults": { + "unit": "percent", + "min": 0, + "max": 100, + "custom": { + "drawStyle": "line", + "lineInterpolation": "linear", + "fillOpacity": 10, + "showPoints": "never" + } + }, + "overrides": [] + } + }, + { + "type": "timeseries", + "id": 12, + "targets": [ + { + "expr": "DCGM_FI_PROF_PIPE_TENSOR_ACTIVE * 100", + "legendFormat": "{{Hostname}}/{{gpu}}", + "refId": "A" + } + ], + "title": "Tensor Pipe Active", + "description": "Real tensor core load (HMMA). For AI/LLM inference it should be ≥20%, otherwise the workload is unoptimized. Cluster-wide — DCGM namespace is the exporter's own namespace, not the workload namespace.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 12, + "y": 7 + }, + "repeatDirection": "h", + "options": { + "legend": { + "displayMode": "table", + "placement": "bottom", + "showLegend": false, + "calcs": [ + "mean", + "max" + ] + }, + "tooltip": { + "mode": "multi", + "sort": "" + } + }, + "fieldConfig": { + "defaults": { + "unit": "percent", + "min": 0, + "max": 100, + "custom": { + "drawStyle": "line", + "lineInterpolation": "linear", + "fillOpacity": 10, + "showPoints": "never" + } + }, + "overrides": [] + } + }, + { + "type": "row", + "collapsed": false, + "title": "Per-GPU ranking", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 15 + }, + "id": 20, + "panels": [] + }, + { + "type": "table", + "id": 21, + "targets": [ + { + "expr": "topk(20, gpu:tensor_saturation:avg5m * 100)", + "instant": true, + "range": false, + "format": "table", + "refId": "A" + } + ], + "title": "Tensor Saturation per GPU (5m avg)", + "description": "Which GPUs are exercising tensor cores and which are not. Sorted descending. Grouped by Hostname/gpu/UUID — DCGM metrics cannot be attributed to workload namespaces.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 0, + "y": 16 + }, + "repeatDirection": "h", + "transformations": [ + { + "id": "organize", + "options": { + "excludeByName": { + "DCGM_FI_DRIVER_VERSION": true, + "Time": true, + "__name__": true, + "cluster": true, + "container": true, + "device": true, + "endpoint": true, + "gpu_driver_version": true, + "instance": true, + "job": true, + "modelName": true, + "namespace": true, + "pci_bus_id": true, + "pod": true, + "prometheus": true, + "service": true, + "tenant": true, + "tier": true, + "uid": true, + "unit": true + }, + "indexByName": { + "Hostname": 0, + "UUID": 2, + "Value": 3, + "gpu": 1 + }, + "renameByName": { + "Hostname": "Node", + "UUID": "UUID", + "Value": "Saturation", + "gpu": "GPU" + } + } + } + ], + "options": { + "frameIndex": 0, + "showHeader": true, + "showTypeIcons": false, + "sortBy": [ + { + "displayName": "Saturation", + "desc": true + } + ], + "footer": { + "show": false, + "reducer": [] + }, + "cellHeight": "sm" + }, + "fieldConfig": { + "defaults": {}, + "overrides": [ + { + "matcher": { + "id": "byName", + "options": "Saturation" + }, + "properties": [ + { + "id": "unit", + "value": "percent" + }, + { + "id": "min", + "value": 0 + }, + { + "id": "max", + "value": 100 + }, + { + "id": "custom.cellOptions", + "value": { + "mode": "gradient", + "type": "gauge" + } + }, + { + "id": "thresholds", + "value": { + "mode": "absolute", + "steps": [ + { + "color": "red" + }, + { + "color": "orange", + "value": 10 + }, + { + "color": "yellow", + "value": 30 + }, + { + "color": "green", + "value": 60 + } + ] + } + } + ] + } + ] + } + }, + { + "type": "table", + "id": 22, + "targets": [ + { + "expr": "topk(20, gpu:util_per_watt:avg5m)", + "instant": true, + "range": false, + "format": "table", + "refId": "A" + } + ], + "title": "Utilization / Watt per GPU (5m avg)", + "description": "How efficiently each GPU spends watts. Low value = poor optimization. Grouped by Hostname/gpu/UUID — DCGM metrics cannot be attributed to workload namespaces.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 12, + "y": 16 + }, + "repeatDirection": "h", + "transformations": [ + { + "id": "organize", + "options": { + "excludeByName": { + "DCGM_FI_DRIVER_VERSION": true, + "Time": true, + "__name__": true, + "cluster": true, + "container": true, + "device": true, + "endpoint": true, + "gpu_driver_version": true, + "instance": true, + "job": true, + "modelName": true, + "namespace": true, + "pci_bus_id": true, + "pod": true, + "prometheus": true, + "service": true, + "tenant": true, + "tier": true, + "uid": true, + "unit": true + }, + "indexByName": { + "Hostname": 0, + "UUID": 2, + "Value": 3, + "gpu": 1 + }, + "renameByName": { + "Hostname": "Node", + "UUID": "UUID", + "Value": "Util/Watt", + "gpu": "GPU" + } + } + } + ], + "options": { + "frameIndex": 0, + "showHeader": true, + "showTypeIcons": false, + "sortBy": [ + { + "displayName": "Util/Watt", + "desc": true + } + ], + "footer": { + "show": false, + "reducer": [] + }, + "cellHeight": "sm" + }, + "fieldConfig": { + "defaults": {}, + "overrides": [ + { + "matcher": { + "id": "byName", + "options": "Util/Watt" + }, + "properties": [ + { + "id": "unit", + "value": "none" + }, + { + "id": "decimals", + "value": 3 + }, + { + "id": "min", + "value": 0 + }, + { + "id": "max", + "value": 1.5 + }, + { + "id": "custom.cellOptions", + "value": { + "mode": "gradient", + "type": "gauge" + } + }, + { + "id": "thresholds", + "value": { + "mode": "absolute", + "steps": [ + { + "color": "red" + }, + { + "color": "yellow", + "value": 0.5 + }, + { + "color": "green", + "value": 1 + } + ] + } + } + ] + } + ] + } + }, + { + "type": "row", + "collapsed": false, + "title": "Throttling", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 24 + }, + "id": 30, + "panels": [] + }, + { + "type": "timeseries", + "id": 31, + "targets": [ + { + "expr": "gpu:power_throttle_fraction:rate5m", + "legendFormat": "{{Hostname}}/{{gpu}}", + "refId": "A" + } + ], + "title": "Power throttle fraction per GPU", + "description": "Fraction of time the GPU was power-throttled. 1.0 = always throttled. Cluster-wide — rule aggregates by (Hostname, gpu, UUID) and drops namespace label.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 0, + "y": 25 + }, + "repeatDirection": "h", + "options": { + "legend": { + "displayMode": "table", + "placement": "bottom", + "showLegend": false, + "calcs": [ + "mean", + "max" + ] + }, + "tooltip": { + "mode": "multi", + "sort": "" + } + }, + "fieldConfig": { + "defaults": { + "unit": "percentunit", + "min": 0, + "max": 1, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "green" + }, + { + "value": 0.05, + "color": "yellow" + }, + { + "value": 0.2, + "color": "red" + } + ] + }, + "custom": { + "drawStyle": "line", + "lineInterpolation": "linear", + "fillOpacity": 25, + "showPoints": "never" + } + }, + "overrides": [] + } + }, + { + "type": "timeseries", + "id": 32, + "targets": [ + { + "expr": "gpu:thermal_throttle_fraction:rate5m", + "legendFormat": "{{Hostname}}/{{gpu}}", + "refId": "A" + } + ], + "title": "Thermal throttle fraction per GPU", + "description": "Fraction of time the GPU was thermal-throttled. Cluster-wide — rule aggregates by (Hostname, gpu, UUID) and drops namespace label.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 12, + "y": 25 + }, + "repeatDirection": "h", + "options": { + "legend": { + "displayMode": "table", + "placement": "bottom", + "showLegend": false, + "calcs": [ + "mean", + "max" + ] + }, + "tooltip": { + "mode": "multi", + "sort": "" + } + }, + "fieldConfig": { + "defaults": { + "unit": "percentunit", + "min": 0, + "max": 1, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "green" + }, + { + "value": 0.05, + "color": "yellow" + }, + { + "value": 0.2, + "color": "red" + } + ] + }, + "custom": { + "drawStyle": "line", + "lineInterpolation": "linear", + "fillOpacity": 25, + "showPoints": "never" + } + }, + "overrides": [] + } + } + ], + "templating": { + "list": [ + { + "type": "datasource", + "name": "ds_prometheus", + "label": "Prometheus", + "skipUrlSync": false, + "query": "prometheus", + "current": { + "selected": false, + "text": "default", + "value": "default" + }, + "multi": false, + "allowCustomValue": true, + "includeAll": false, + "regex": "", + "auto": false, + "auto_min": "10s", + "auto_count": 30 + } + ] + }, + "annotations": {} +} diff --git a/dashboards/gpu/gpu-fleet.json b/dashboards/gpu/gpu-fleet.json new file mode 100644 index 00000000..60c12bfb --- /dev/null +++ b/dashboards/gpu/gpu-fleet.json @@ -0,0 +1,1157 @@ +{ + "uid": "gpu-fleet", + "title": "GPU Fleet Overview", + "description": "Cluster-wide GPU inventory, capacity, utilization and health — admin view across the whole fleet.", + "tags": [ + "gpu", + "fleet", + "admin" + ], + "timezone": "browser", + "editable": true, + "graphTooltip": 1, + "time": { + "from": "now-6h", + "to": "now" + }, + "fiscalYearStartMonth": 0, + "schemaVersion": 42, + "panels": [ + { + "type": "row", + "collapsed": false, + "title": "Cluster snapshot", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 0 + }, + "id": 1, + "panels": [] + }, + { + "type": "stat", + "id": 2, + "targets": [ + { + "expr": "cluster:gpu_count:total", + "refId": "A" + } + ], + "title": "Total GPUs", + "description": "Physical GPUs discovered by DCGM across all nodes.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 4, + "w": 4, + "x": 0, + "y": 1 + }, + "repeatDirection": "h", + "options": { + "graphMode": "none", + "colorMode": "value", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "short", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "blue" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "stat", + "id": 3, + "targets": [ + { + "expr": "cluster:gpu_count:allocated", + "refId": "A" + } + ], + "title": "Allocated", + "description": "Sum of kube GPU requests across all pods. Billable view — includes Pending pods.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 4, + "w": 4, + "x": 4, + "y": 1 + }, + "repeatDirection": "h", + "options": { + "graphMode": "none", + "colorMode": "background", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "short", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "blue" + }, + { + "value": 1, + "color": "green" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "stat", + "id": 4, + "targets": [ + { + "expr": "cluster:gpu_count:free", + "refId": "A" + } + ], + "title": "Free", + "description": "Unallocated capacity available for new tenants.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 4, + "w": 4, + "x": 8, + "y": 1 + }, + "repeatDirection": "h", + "options": { + "graphMode": "none", + "colorMode": "background", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "short", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "red" + }, + { + "value": 1, + "color": "yellow" + }, + { + "value": 2, + "color": "green" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "stat", + "id": 5, + "targets": [ + { + "expr": "count(count by (Hostname) (DCGM_FI_DEV_GPU_UTIL))", + "refId": "A" + } + ], + "title": "Nodes with GPU", + "description": "Nodes reporting at least one GPU to DCGM. Counted via DCGM_FI_DEV_GPU_UTIL because kube-state-metrics does not expose node labels by default.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 4, + "w": 4, + "x": 12, + "y": 1 + }, + "repeatDirection": "h", + "options": { + "graphMode": "none", + "colorMode": "value", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "short", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "blue" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "stat", + "id": 6, + "targets": [ + { + "expr": "cluster:gpu_util:avg", + "refId": "A" + } + ], + "title": "Average utilization", + "description": "Legacy NVML utilization across all tenant GPUs.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 4, + "w": 4, + "x": 16, + "y": 1 + }, + "repeatDirection": "h", + "options": { + "graphMode": "area", + "colorMode": "value", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "percent", + "min": 0, + "max": 100, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "blue" + }, + { + "value": 30, + "color": "green" + }, + { + "value": 80, + "color": "orange" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "stat", + "id": 7, + "targets": [ + { + "expr": "cluster:gpu_power_watts:sum", + "refId": "A" + } + ], + "title": "Total power", + "description": "Live power draw summed across the fleet.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 4, + "w": 4, + "x": 20, + "y": 1 + }, + "repeatDirection": "h", + "options": { + "graphMode": "area", + "colorMode": "value", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "watt", + "decimals": 0, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "green" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "row", + "collapsed": false, + "title": "Per-node GPU activity", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 5 + }, + "id": 10, + "panels": [] + }, + { + "type": "bargauge", + "id": 11, + "targets": [ + { + "expr": "avg by (Hostname) (DCGM_FI_DEV_GPU_UTIL)", + "legendFormat": "{{Hostname}}", + "refId": "A" + } + ], + "title": "GPU utilization per node", + "description": "Average NVML utilization per node — spot idle or saturated hosts at a glance.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 0, + "y": 6 + }, + "repeatDirection": "h", + "options": { + "displayMode": "gradient", + "valueMode": "color", + "namePlacement": "auto", + "showUnfilled": true, + "sizing": "auto", + "minVizWidth": 8, + "minVizHeight": 16, + "legend": { + "displayMode": "list", + "placement": "bottom", + "showLegend": false, + "calcs": [] + }, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "maxVizHeight": 300, + "orientation": "horizontal" + }, + "fieldConfig": { + "defaults": { + "unit": "percent", + "min": 0, + "max": 100, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "blue" + }, + { + "value": 30, + "color": "green" + }, + { + "value": 80, + "color": "orange" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "bargauge", + "id": 12, + "targets": [ + { + "expr": "sum by (Hostname) (DCGM_FI_DEV_POWER_USAGE) / clamp_min(sum by (Hostname) (DCGM_FI_DEV_POWER_MGMT_LIMIT), 1) * 100", + "legendFormat": "{{Hostname}}", + "refId": "A" + } + ], + "title": "Power draw per node (% of TDP)", + "description": "GPU power usage as a fraction of the node's combined TDP cap. Works across GPU SKUs without per-model thresholds. Requires DCGM_FI_DEV_POWER_MGMT_LIMIT from custom DCGM CSV (see packages/system/gpu-operator/examples/dcgm-custom-metrics.yaml).", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 12, + "y": 6 + }, + "repeatDirection": "h", + "options": { + "displayMode": "gradient", + "valueMode": "color", + "namePlacement": "auto", + "showUnfilled": true, + "sizing": "auto", + "minVizWidth": 8, + "minVizHeight": 16, + "legend": { + "displayMode": "list", + "placement": "bottom", + "showLegend": false, + "calcs": [] + }, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "maxVizHeight": 300, + "orientation": "horizontal" + }, + "fieldConfig": { + "defaults": { + "unit": "percent", + "min": 0, + "max": 100, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "green" + }, + { + "value": 60, + "color": "yellow" + }, + { + "value": 80, + "color": "orange" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "row", + "collapsed": false, + "title": "Usage trends", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 14 + }, + "id": 20, + "panels": [] + }, + { + "type": "timeseries", + "id": 21, + "targets": [ + { + "expr": "avg by (Hostname) (DCGM_FI_DEV_GPU_UTIL)", + "legendFormat": "{{Hostname}}", + "refId": "A" + } + ], + "title": "GPU utilization per node (trend)", + "description": "Per-node NVML utilization over time — shows shifting workload across hosts.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 0, + "y": 15 + }, + "repeatDirection": "h", + "options": { + "legend": { + "displayMode": "table", + "placement": "bottom", + "showLegend": false, + "calcs": [ + "mean", + "max" + ] + }, + "tooltip": { + "mode": "multi", + "sort": "" + } + }, + "fieldConfig": { + "defaults": { + "unit": "percent", + "min": 0, + "max": 100, + "custom": { + "drawStyle": "line", + "lineInterpolation": "linear", + "fillOpacity": 20, + "showPoints": "never" + } + }, + "overrides": [] + } + }, + { + "type": "timeseries", + "id": 22, + "targets": [ + { + "expr": "sum by (Hostname) (DCGM_FI_DEV_POWER_USAGE)", + "legendFormat": "{{Hostname}}", + "refId": "A" + } + ], + "title": "Power draw per node (trend)", + "description": "Per-node GPU power consumption over time.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 12, + "y": 15 + }, + "repeatDirection": "h", + "options": { + "legend": { + "displayMode": "table", + "placement": "bottom", + "showLegend": false, + "calcs": [ + "mean", + "max" + ] + }, + "tooltip": { + "mode": "multi", + "sort": "" + } + }, + "fieldConfig": { + "defaults": { + "unit": "watt", + "custom": { + "drawStyle": "line", + "lineInterpolation": "linear", + "fillOpacity": 20, + "showPoints": "never" + } + }, + "overrides": [] + } + }, + { + "type": "row", + "collapsed": false, + "title": "Health", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 23 + }, + "id": 30, + "panels": [] + }, + { + "type": "stat", + "id": 31, + "targets": [ + { + "expr": "count((max by (Hostname, gpu) (DCGM_FI_DEV_XID_ERRORS)) \u003e 0) or vector(0)", + "refId": "A" + } + ], + "title": "GPUs with XID errors", + "description": "Count of GPUs reporting a non-zero XID code. XID = NVIDIA hardware/driver error — any non-zero value needs investigation.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 5, + "w": 6, + "x": 0, + "y": 24 + }, + "repeatDirection": "h", + "options": { + "graphMode": "none", + "colorMode": "background", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "short", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "green" + }, + { + "value": 1, + "color": "red" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "stat", + "id": 32, + "targets": [ + { + "expr": "avg(gpu:power_throttle_fraction:rate5m)", + "refId": "A" + } + ], + "title": "Power throttling (cluster avg)", + "description": "Fraction of time the fleet spent power-throttled. \u003e5% = billed FLOPS are being underdelivered.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 5, + "w": 6, + "x": 6, + "y": 24 + }, + "repeatDirection": "h", + "options": { + "graphMode": "none", + "colorMode": "background", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "percentunit", + "decimals": 2, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "green" + }, + { + "value": 0.05, + "color": "yellow" + }, + { + "value": 0.2, + "color": "red" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "stat", + "id": 33, + "targets": [ + { + "expr": "avg(gpu:thermal_throttle_fraction:rate5m)", + "refId": "A" + } + ], + "title": "Thermal throttling (cluster avg)", + "description": "Fraction of time the fleet spent thermal-throttled. Indicates cooling/datacenter issues.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 5, + "w": 6, + "x": 12, + "y": 24 + }, + "repeatDirection": "h", + "options": { + "graphMode": "none", + "colorMode": "background", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "percentunit", + "decimals": 2, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "green" + }, + { + "value": 0.05, + "color": "yellow" + }, + { + "value": 0.2, + "color": "red" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "stat", + "id": 34, + "targets": [ + { + "expr": "max(DCGM_FI_DEV_GPU_TEMP)", + "refId": "A" + } + ], + "title": "Max GPU temperature", + "description": "Hottest GPU in the fleet right now. Adjust thresholds to match your GPU model specifications.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 5, + "w": 6, + "x": 18, + "y": 24 + }, + "repeatDirection": "h", + "options": { + "graphMode": "none", + "colorMode": "background", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "celsius", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "green" + }, + { + "value": 75, + "color": "yellow" + }, + { + "value": 85, + "color": "red" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "timeseries", + "id": 35, + "targets": [ + { + "expr": "DCGM_FI_DEV_GPU_TEMP", + "legendFormat": "{{Hostname}}/{{gpu}}", + "refId": "A" + } + ], + "title": "Temperature per GPU", + "description": "Per-GPU temperature trace. Sustained red zone = cooling issue or aggressive workload.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 0, + "y": 29 + }, + "repeatDirection": "h", + "options": { + "legend": { + "displayMode": "table", + "placement": "bottom", + "showLegend": false, + "calcs": [ + "mean", + "max" + ] + }, + "tooltip": { + "mode": "multi", + "sort": "" + } + }, + "fieldConfig": { + "defaults": { + "unit": "celsius", + "min": 20, + "max": 100, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "green" + }, + { + "value": 75, + "color": "yellow" + }, + { + "value": 85, + "color": "red" + } + ] + }, + "custom": { + "drawStyle": "line", + "lineInterpolation": "linear", + "fillOpacity": 10, + "showPoints": "never" + } + }, + "overrides": [] + } + }, + { + "type": "timeseries", + "id": 36, + "targets": [ + { + "expr": "gpu:power_throttle_fraction:rate5m", + "legendFormat": "power {{Hostname}}/{{gpu}}", + "refId": "A" + }, + { + "expr": "gpu:thermal_throttle_fraction:rate5m", + "legendFormat": "thermal {{Hostname}}/{{gpu}}", + "refId": "B" + } + ], + "title": "Throttle fraction per GPU", + "description": "Red = power throttle, blue = thermal throttle. Both on same chart to compare causes.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 12, + "y": 29 + }, + "repeatDirection": "h", + "options": { + "legend": { + "displayMode": "table", + "placement": "bottom", + "showLegend": false, + "calcs": [ + "mean", + "max" + ] + }, + "tooltip": { + "mode": "multi", + "sort": "" + } + }, + "fieldConfig": { + "defaults": { + "unit": "percentunit", + "min": 0, + "max": 1, + "custom": { + "drawStyle": "line", + "lineInterpolation": "linear", + "fillOpacity": 25, + "showPoints": "never" + } + }, + "overrides": [] + } + }, + { + "type": "row", + "collapsed": false, + "title": "Hardware inventory", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 37 + }, + "id": 40, + "panels": [] + }, + { + "type": "table", + "id": 41, + "targets": [ + { + "expr": "group by (Hostname, modelName, gpu, UUID) (DCGM_FI_DEV_GPU_UTIL)", + "instant": true, + "range": false, + "format": "table", + "refId": "A" + } + ], + "title": "GPU inventory", + "description": "One row per physical GPU. Model / index / UUID sourced from DCGM labels.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 6, + "w": 24, + "x": 0, + "y": 38 + }, + "repeatDirection": "h", + "transformations": [ + { + "id": "organize", + "options": { + "excludeByName": { + "Time": true, + "Value": true, + "__name__": true, + "cluster": true, + "container": true, + "device": true, + "endpoint": true, + "instance": true, + "job": true, + "pci_bus_id": true, + "prometheus": true, + "service": true, + "tenant": true, + "tier": true, + "uid": true, + "unit": true + }, + "indexByName": { + "Hostname": 0, + "UUID": 3, + "gpu": 2, + "modelName": 1 + }, + "renameByName": { + "Hostname": "Node", + "UUID": "UUID", + "gpu": "Index", + "modelName": "GPU Model" + } + } + } + ], + "options": { + "frameIndex": 0, + "showHeader": true, + "showTypeIcons": false, + "footer": { + "show": false, + "reducer": [] + }, + "cellHeight": "sm" + } + } + ], + "templating": { + "list": [ + { + "type": "datasource", + "name": "ds_prometheus", + "label": "Prometheus", + "skipUrlSync": false, + "query": "prometheus", + "current": { + "selected": false, + "text": "default", + "value": "default" + }, + "multi": false, + "allowCustomValue": true, + "includeAll": false, + "regex": "", + "auto": false, + "auto_min": "10s", + "auto_count": 30 + } + ] + }, + "annotations": {} +} diff --git a/dashboards/gpu/gpu-performance.json b/dashboards/gpu/gpu-performance.json new file mode 100644 index 00000000..aed9c69a --- /dev/null +++ b/dashboards/gpu/gpu-performance.json @@ -0,0 +1,957 @@ +{ + "uid": "gpu-performance", + "title": "GPU Performance", + "tags": [ + "gpu", + "dcgm" + ], + "timezone": "browser", + "editable": true, + "graphTooltip": 1, + "time": { + "from": "now-1h", + "to": "now" + }, + "fiscalYearStartMonth": 0, + "schemaVersion": 42, + "panels": [ + { + "type": "row", + "collapsed": false, + "title": "Overview", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 0 + }, + "id": 1, + "panels": [] + }, + { + "type": "stat", + "id": 2, + "targets": [ + { + "expr": "cluster:gpu_count:total", + "refId": "A" + } + ], + "title": "Total GPUs", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 4, + "w": 6, + "x": 0, + "y": 1 + }, + "repeatDirection": "h", + "options": { + "graphMode": "none", + "colorMode": "value", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "short", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "blue" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "stat", + "id": 3, + "targets": [ + { + "expr": "cluster:gpu_count:allocated", + "refId": "A" + } + ], + "title": "Allocated", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 4, + "w": 6, + "x": 6, + "y": 1 + }, + "repeatDirection": "h", + "options": { + "graphMode": "none", + "colorMode": "value", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "short", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "green" + }, + { + "value": 1, + "color": "yellow" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "stat", + "id": 4, + "targets": [ + { + "expr": "cluster:gpu_util:avg", + "refId": "A" + } + ], + "title": "Average utilization", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 4, + "w": 6, + "x": 12, + "y": 1 + }, + "repeatDirection": "h", + "options": { + "graphMode": "area", + "colorMode": "value", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "percent", + "min": 0, + "max": 100, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "blue" + }, + { + "value": 30, + "color": "green" + }, + { + "value": 80, + "color": "orange" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "stat", + "id": 5, + "targets": [ + { + "expr": "cluster:gpu_power_watts:sum", + "refId": "A" + } + ], + "title": "Power draw", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 4, + "w": 6, + "x": 18, + "y": 1 + }, + "repeatDirection": "h", + "options": { + "graphMode": "area", + "colorMode": "value", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "watt", + "decimals": 0, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "green" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "row", + "collapsed": false, + "title": "Utilization", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 5 + }, + "id": 10, + "panels": [] + }, + { + "type": "timeseries", + "id": 11, + "targets": [ + { + "expr": "DCGM_FI_DEV_GPU_UTIL{Hostname=~\"$Hostname\"}", + "legendFormat": "{{Hostname}}/{{gpu}} {{pod}}", + "refId": "A" + } + ], + "title": "GPU Utilization (NVML)", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 0, + "y": 6 + }, + "repeatDirection": "h", + "options": { + "legend": { + "displayMode": "table", + "placement": "bottom", + "showLegend": false, + "calcs": [ + "mean", + "max" + ] + }, + "tooltip": { + "mode": "multi", + "sort": "" + } + }, + "fieldConfig": { + "defaults": { + "unit": "percent", + "min": 0, + "max": 100, + "custom": { + "drawStyle": "line", + "lineInterpolation": "linear", + "fillOpacity": 10, + "showPoints": "never", + "spanNulls": false + } + }, + "overrides": [] + } + }, + { + "type": "timeseries", + "id": 12, + "targets": [ + { + "expr": "DCGM_FI_PROF_PIPE_TENSOR_ACTIVE{Hostname=~\"$Hostname\"} * 100", + "legendFormat": "{{Hostname}}/{{gpu}} {{pod}}", + "refId": "A" + } + ], + "title": "Tensor Pipe Active (realistic load for LLM/AI)", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 12, + "y": 6 + }, + "repeatDirection": "h", + "options": { + "legend": { + "displayMode": "table", + "placement": "bottom", + "showLegend": false, + "calcs": [ + "mean", + "max" + ] + }, + "tooltip": { + "mode": "multi", + "sort": "" + } + }, + "fieldConfig": { + "defaults": { + "unit": "percent", + "min": 0, + "max": 100, + "custom": { + "drawStyle": "line", + "lineInterpolation": "linear", + "fillOpacity": 10, + "showPoints": "never", + "spanNulls": false + } + }, + "overrides": [] + } + }, + { + "type": "timeseries", + "id": 13, + "targets": [ + { + "expr": "DCGM_FI_PROF_GR_ENGINE_ACTIVE{Hostname=~\"$Hostname\"} * 100", + "legendFormat": "{{Hostname}}/{{gpu}} {{pod}}", + "refId": "A" + } + ], + "title": "Graphics Engine Active", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 0, + "y": 14 + }, + "repeatDirection": "h", + "options": { + "legend": { + "displayMode": "table", + "placement": "bottom", + "showLegend": false, + "calcs": [ + "mean", + "max" + ] + }, + "tooltip": { + "mode": "multi", + "sort": "" + } + }, + "fieldConfig": { + "defaults": { + "unit": "percent", + "min": 0, + "max": 100, + "custom": { + "drawStyle": "line", + "lineInterpolation": "linear", + "fillOpacity": 10, + "showPoints": "never", + "spanNulls": false + } + }, + "overrides": [] + } + }, + { + "type": "timeseries", + "id": 14, + "targets": [ + { + "expr": "DCGM_FI_DEV_MEM_COPY_UTIL{Hostname=~\"$Hostname\"}", + "legendFormat": "{{Hostname}}/{{gpu}} {{pod}}", + "refId": "A" + } + ], + "title": "Memory Copy Utilization", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 12, + "y": 14 + }, + "repeatDirection": "h", + "options": { + "legend": { + "displayMode": "table", + "placement": "bottom", + "showLegend": false, + "calcs": [ + "mean", + "max" + ] + }, + "tooltip": { + "mode": "multi", + "sort": "" + } + }, + "fieldConfig": { + "defaults": { + "unit": "percent", + "min": 0, + "max": 100, + "custom": { + "drawStyle": "line", + "lineInterpolation": "linear", + "fillOpacity": 10, + "showPoints": "never", + "spanNulls": false + } + }, + "overrides": [] + } + }, + { + "type": "row", + "collapsed": false, + "title": "Memory", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 22 + }, + "id": 20, + "panels": [] + }, + { + "type": "timeseries", + "id": 21, + "targets": [ + { + "expr": "DCGM_FI_DEV_FB_USED{Hostname=~\"$Hostname\"} * 1048576", + "legendFormat": "{{Hostname}}/{{gpu}} {{pod}}", + "refId": "A" + } + ], + "title": "VRAM Used", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 0, + "y": 23 + }, + "repeatDirection": "h", + "options": { + "legend": { + "displayMode": "table", + "placement": "bottom", + "showLegend": false, + "calcs": [ + "mean", + "max" + ] + }, + "tooltip": { + "mode": "multi", + "sort": "" + } + }, + "fieldConfig": { + "defaults": { + "unit": "bytes", + "custom": { + "drawStyle": "line", + "lineInterpolation": "linear", + "fillOpacity": 25, + "showPoints": "never" + } + }, + "overrides": [] + } + }, + { + "type": "timeseries", + "id": 22, + "targets": [ + { + "expr": "DCGM_FI_DEV_FB_FREE{Hostname=~\"$Hostname\"} * 1048576", + "legendFormat": "{{Hostname}}/{{gpu}}", + "refId": "A" + } + ], + "title": "VRAM Free", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 12, + "y": 23 + }, + "repeatDirection": "h", + "options": { + "legend": { + "displayMode": "table", + "placement": "bottom", + "showLegend": false, + "calcs": [ + "mean", + "min" + ] + }, + "tooltip": { + "mode": "multi", + "sort": "" + } + }, + "fieldConfig": { + "defaults": { + "unit": "bytes", + "custom": { + "drawStyle": "line", + "lineInterpolation": "linear", + "fillOpacity": 10, + "showPoints": "never" + } + }, + "overrides": [] + } + }, + { + "type": "row", + "collapsed": false, + "title": "Power \u0026 Temperature", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 31 + }, + "id": 30, + "panels": [] + }, + { + "type": "timeseries", + "id": 31, + "targets": [ + { + "expr": "DCGM_FI_DEV_POWER_USAGE{Hostname=~\"$Hostname\"}", + "legendFormat": "{{Hostname}}/{{gpu}}", + "refId": "A" + } + ], + "title": "Power Usage per GPU", + "description": "Hardware-level metric — shown cluster-wide regardless of namespace filter.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 0, + "y": 32 + }, + "repeatDirection": "h", + "options": { + "legend": { + "displayMode": "table", + "placement": "bottom", + "showLegend": false, + "calcs": [ + "mean", + "max" + ] + }, + "tooltip": { + "mode": "multi", + "sort": "" + } + }, + "fieldConfig": { + "defaults": { + "unit": "watt", + "custom": { + "drawStyle": "line", + "lineInterpolation": "linear", + "fillOpacity": 10, + "showPoints": "never" + } + }, + "overrides": [] + } + }, + { + "type": "timeseries", + "id": 32, + "targets": [ + { + "expr": "DCGM_FI_DEV_GPU_TEMP{Hostname=~\"$Hostname\"}", + "legendFormat": "{{Hostname}}/{{gpu}}", + "refId": "A" + } + ], + "title": "GPU Temperature", + "description": "Hardware-level metric — shown cluster-wide regardless of namespace filter.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 12, + "y": 32 + }, + "repeatDirection": "h", + "options": { + "legend": { + "displayMode": "table", + "placement": "bottom", + "showLegend": false, + "calcs": [ + "mean", + "max" + ] + }, + "tooltip": { + "mode": "multi", + "sort": "" + } + }, + "fieldConfig": { + "defaults": { + "unit": "celsius", + "min": 20, + "max": 100, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "green" + }, + { + "value": 75, + "color": "yellow" + }, + { + "value": 85, + "color": "red" + } + ] + }, + "custom": { + "drawStyle": "line", + "lineInterpolation": "linear", + "fillOpacity": 10, + "showPoints": "never" + } + }, + "overrides": [] + } + }, + { + "type": "row", + "collapsed": false, + "title": "Health", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 40 + }, + "id": 40, + "panels": [] + }, + { + "type": "stat", + "id": 41, + "targets": [ + { + "expr": "max by (Hostname, gpu) (DCGM_FI_DEV_XID_ERRORS{Hostname=~\"$Hostname\"})", + "legendFormat": "{{Hostname}}/{{gpu}}", + "refId": "A" + } + ], + "title": "XID errors (latest)", + "description": "Hardware-level metric — shown cluster-wide regardless of namespace filter.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 5, + "w": 8, + "x": 0, + "y": 41 + }, + "repeatDirection": "h", + "options": { + "graphMode": "none", + "colorMode": "background", + "justifyMode": "auto", + "textMode": "value_and_name", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "short", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "green" + }, + { + "value": 1, + "color": "red" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "timeseries", + "id": 42, + "targets": [ + { + "expr": "rate(DCGM_FI_DEV_POWER_VIOLATION{Hostname=~\"$Hostname\"}[5m])", + "legendFormat": "{{Hostname}}/{{gpu}}", + "refId": "A" + } + ], + "title": "Power Violation (µs/s throttled due to power)", + "description": "Hardware-level metric — shown cluster-wide regardless of namespace filter.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 5, + "w": 8, + "x": 8, + "y": 41 + }, + "repeatDirection": "h", + "options": { + "legend": { + "displayMode": "list", + "placement": "bottom", + "showLegend": false, + "calcs": [] + }, + "tooltip": { + "mode": "multi", + "sort": "" + } + }, + "fieldConfig": { + "defaults": { + "unit": "µs", + "custom": { + "drawStyle": "line", + "lineInterpolation": "linear", + "fillOpacity": 10, + "showPoints": "never" + } + }, + "overrides": [] + } + }, + { + "type": "timeseries", + "id": 43, + "targets": [ + { + "expr": "rate(DCGM_FI_DEV_THERMAL_VIOLATION{Hostname=~\"$Hostname\"}[5m])", + "legendFormat": "{{Hostname}}/{{gpu}}", + "refId": "A" + } + ], + "title": "Thermal Violation (µs/s throttled due to thermals)", + "description": "Hardware-level metric — shown cluster-wide regardless of namespace filter.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 5, + "w": 8, + "x": 16, + "y": 41 + }, + "repeatDirection": "h", + "options": { + "legend": { + "displayMode": "list", + "placement": "bottom", + "showLegend": false, + "calcs": [] + }, + "tooltip": { + "mode": "multi", + "sort": "" + } + }, + "fieldConfig": { + "defaults": { + "unit": "µs", + "custom": { + "drawStyle": "line", + "lineInterpolation": "linear", + "fillOpacity": 10, + "showPoints": "never" + } + }, + "overrides": [] + } + } + ], + "templating": { + "list": [ + { + "type": "datasource", + "name": "ds_prometheus", + "label": "Prometheus", + "skipUrlSync": false, + "query": "prometheus", + "current": { + "selected": false, + "text": "default", + "value": "default" + }, + "multi": false, + "allowCustomValue": true, + "includeAll": false, + "regex": "", + "auto": false, + "auto_min": "10s", + "auto_count": 30 + }, + { + "type": "query", + "name": "Hostname", + "label": "Host", + "skipUrlSync": false, + "query": "label_values(DCGM_FI_DEV_GPU_UTIL, Hostname)", + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "current": { + "selected": true, + "text": "All", + "value": "$__all" + }, + "multi": true, + "allowCustomValue": true, + "refresh": 2, + "sort": 1, + "includeAll": true, + "auto": false, + "auto_min": "10s", + "auto_count": 30 + } + ] + }, + "annotations": {} +} diff --git a/dashboards/gpu/gpu-quotas.json b/dashboards/gpu/gpu-quotas.json new file mode 100644 index 00000000..865fb95c --- /dev/null +++ b/dashboards/gpu/gpu-quotas.json @@ -0,0 +1,637 @@ +{ + "uid": "gpu-quotas", + "title": "GPU Quotas \u0026 Allocation", + "tags": [ + "gpu", + "quotas" + ], + "timezone": "browser", + "editable": true, + "graphTooltip": 1, + "time": { + "from": "now-6h", + "to": "now" + }, + "fiscalYearStartMonth": 0, + "schemaVersion": 42, + "panels": [ + { + "type": "row", + "collapsed": false, + "title": "Allocation overview", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 0 + }, + "id": 1, + "panels": [] + }, + { + "type": "stat", + "id": 2, + "targets": [ + { + "expr": "sum(kube_node_status_allocatable{resource=\"nvidia_com_gpu\"})", + "refId": "A" + } + ], + "title": "GPU allocatable", + "description": "Total GPU capacity the cluster can schedule to pods.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 4, + "w": 6, + "x": 0, + "y": 1 + }, + "repeatDirection": "h", + "options": { + "graphMode": "none", + "colorMode": "value", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "short", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "blue" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "stat", + "id": 3, + "targets": [ + { + "expr": "cluster:gpu_count:allocated", + "refId": "A" + } + ], + "title": "GPU requested", + "description": "Sum of GPU requests across all pods cluster-wide, including system namespaces.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 4, + "w": 6, + "x": 6, + "y": 1 + }, + "repeatDirection": "h", + "options": { + "graphMode": "none", + "colorMode": "value", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "short", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "green" + }, + { + "value": 1, + "color": "yellow" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "gauge", + "id": 4, + "targets": [ + { + "expr": "cluster:gpu_count:allocated / sum(kube_node_status_allocatable{resource=\"nvidia_com_gpu\"}) * 100", + "refId": "A" + } + ], + "title": "Allocation ratio", + "description": "Percentage of allocatable GPUs currently requested by pods.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 4, + "w": 6, + "x": 12, + "y": 1 + }, + "repeatDirection": "h", + "options": { + "showThresholdLabels": false, + "showThresholdMarkers": true, + "sizing": "auto", + "minVizWidth": 75, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "minVizHeight": 75, + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "percent", + "min": 0, + "max": 100, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "blue" + }, + { + "value": 40, + "color": "green" + }, + { + "value": 80, + "color": "yellow" + }, + { + "value": 95, + "color": "red" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "stat", + "id": 5, + "targets": [ + { + "expr": "count((sum by (namespace, pod) (kube_pod_container_resource_requests{resource=\"nvidia_com_gpu\"}) \u003e 0) * on(namespace, pod) group_left() (kube_pod_status_phase{phase=\"Pending\"} == 1)) or vector(0)", + "refId": "A" + } + ], + "title": "Pending pods (GPU)", + "description": "Pods requesting GPUs that are stuck in Pending state — indicates capacity shortage.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 4, + "w": 6, + "x": 18, + "y": 1 + }, + "repeatDirection": "h", + "options": { + "graphMode": "none", + "colorMode": "background", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "short", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "green" + }, + { + "value": 1, + "color": "red" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "row", + "collapsed": false, + "title": "Per namespace", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 5 + }, + "id": 10, + "panels": [] + }, + { + "type": "bargauge", + "id": 11, + "targets": [ + { + "expr": "sum by (namespace) (namespace:gpu_count:allocated{namespace=~\"$namespace\"})", + "legendFormat": "{{namespace}}", + "refId": "A" + } + ], + "title": "GPU requested per namespace", + "description": "GPU allocation breakdown by namespace — spot top consumers at a glance.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 0, + "y": 6 + }, + "repeatDirection": "h", + "options": { + "displayMode": "gradient", + "valueMode": "color", + "namePlacement": "auto", + "showUnfilled": true, + "sizing": "auto", + "minVizWidth": 8, + "minVizHeight": 16, + "legend": { + "displayMode": "list", + "placement": "bottom", + "showLegend": false, + "calcs": [] + }, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "maxVizHeight": 300, + "orientation": "horizontal" + }, + "fieldConfig": { + "defaults": { + "unit": "short", + "min": 0, + "max": 8, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "green" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "timeseries", + "id": 12, + "targets": [ + { + "expr": "sum(kube_node_status_allocatable{resource=\"nvidia_com_gpu\"})", + "legendFormat": "Allocatable (total)", + "refId": "A" + }, + { + "expr": "sum(namespace:gpu_count:allocated{namespace=~\"$namespace\"})", + "legendFormat": "Requested", + "refId": "B" + } + ], + "title": "GPU allocated over time", + "description": "Requested vs allocatable GPUs over time — shows allocation pressure trends.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 12, + "y": 6 + }, + "repeatDirection": "h", + "options": { + "legend": { + "displayMode": "list", + "placement": "bottom", + "showLegend": false, + "calcs": [] + }, + "tooltip": { + "mode": "multi", + "sort": "" + } + }, + "fieldConfig": { + "defaults": { + "unit": "short", + "custom": { + "drawStyle": "line", + "lineInterpolation": "stepAfter", + "fillOpacity": 10, + "showPoints": "never" + } + }, + "overrides": [ + { + "matcher": { + "id": "byName", + "options": "Allocatable (total)" + }, + "properties": [ + { + "id": "color", + "value": { + "fixedColor": "blue", + "mode": "fixed" + } + } + ] + } + ] + } + }, + { + "type": "row", + "collapsed": false, + "title": "Pods with GPU", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 14 + }, + "id": 20, + "panels": [] + }, + { + "type": "table", + "id": 21, + "targets": [ + { + "expr": "kube_pod_container_resource_requests{resource=\"nvidia_com_gpu\", namespace=~\"$namespace\"} * on(namespace, pod) group_left(phase) (kube_pod_status_phase{phase=~\"Running|Pending\"} == 1)", + "instant": true, + "range": false, + "format": "table", + "refId": "requested" + }, + { + "expr": "kube_pod_container_resource_limits{resource=\"nvidia_com_gpu\", namespace=~\"$namespace\"} * on(namespace, pod) group_left() (kube_pod_status_phase{phase=~\"Running|Pending\"} == 1)", + "instant": true, + "range": false, + "format": "table", + "refId": "limits" + } + ], + "title": "Pods requesting GPU", + "description": "Per-pod GPU requests and limits with scheduling status — Running or Pending.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 10, + "w": 24, + "x": 0, + "y": 15 + }, + "repeatDirection": "h", + "transformations": [ + { + "id": "joinByField", + "options": { + "byField": "pod", + "mode": "outer" + } + }, + { + "id": "organize", + "options": { + "excludeByName": { + "Time": true, + "Time 1": true, + "Time 2": true, + "__name__": true, + "__name__ 1": true, + "__name__ 2": true, + "cluster": true, + "cluster 1": true, + "cluster 2": true, + "container 2": true, + "endpoint": true, + "endpoint 1": true, + "endpoint 2": true, + "instance": true, + "instance 1": true, + "instance 2": true, + "job": true, + "job 1": true, + "job 2": true, + "namespace 2": true, + "node 2": true, + "prometheus": true, + "prometheus 1": true, + "prometheus 2": true, + "resource": true, + "resource 1": true, + "resource 2": true, + "service": true, + "service 1": true, + "service 2": true, + "tenant": true, + "tenant 1": true, + "tenant 2": true, + "tier": true, + "tier 1": true, + "tier 2": true, + "uid": true, + "uid 1": true, + "uid 2": true, + "unit": true, + "unit 1": true, + "unit 2": true + }, + "indexByName": { + "Value #limits": 5, + "Value #requested": 4, + "container 1": 2, + "namespace 1": 0, + "node 1": 3, + "phase": 6, + "pod": 1 + }, + "renameByName": { + "Value #limits": "Limit", + "Value #requested": "Req", + "container 1": "Container", + "namespace 1": "Namespace", + "node 1": "Node", + "phase": "Status" + } + } + } + ], + "options": { + "frameIndex": 0, + "showHeader": true, + "showTypeIcons": false, + "footer": { + "show": false, + "reducer": [] + }, + "cellHeight": "sm" + }, + "fieldConfig": { + "defaults": {}, + "overrides": [ + { + "matcher": { + "id": "byName", + "options": "Status" + }, + "properties": [ + { + "id": "custom.cellOptions", + "value": { + "mode": "basic", + "type": "color-background" + } + }, + { + "id": "mappings", + "value": [ + { + "options": { + "Failed": { + "color": "red", + "index": 2 + }, + "Pending": { + "color": "orange", + "index": 1 + }, + "Running": { + "color": "green", + "index": 0 + } + }, + "type": "value" + } + ] + } + ] + } + ] + } + } + ], + "templating": { + "list": [ + { + "type": "datasource", + "name": "ds_prometheus", + "label": "Prometheus", + "skipUrlSync": false, + "query": "prometheus", + "current": { + "selected": false, + "text": "default", + "value": "default" + }, + "multi": false, + "allowCustomValue": true, + "includeAll": false, + "regex": "", + "auto": false, + "auto_min": "10s", + "auto_count": 30 + }, + { + "type": "query", + "name": "namespace", + "label": "Namespace", + "skipUrlSync": false, + "query": "label_values(kube_pod_container_resource_requests{resource=\"nvidia_com_gpu\"}, namespace)", + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "current": { + "selected": true, + "text": "All", + "value": "$__all" + }, + "multi": true, + "allowCustomValue": true, + "refresh": 2, + "sort": 1, + "includeAll": true, + "auto": false, + "auto_min": "10s", + "auto_count": 30 + } + ] + }, + "annotations": {} +} diff --git a/dashboards/gpu/gpu-tenants.json b/dashboards/gpu/gpu-tenants.json new file mode 100644 index 00000000..7fef321d --- /dev/null +++ b/dashboards/gpu/gpu-tenants.json @@ -0,0 +1,1098 @@ +{ + "uid": "gpu-tenants", + "title": "GPU Tenant Usage", + "description": "Per-tenant GPU consumption: live allocation, utilization, tensor saturation, power, and 24h GPU-hours / kWh — admin view across all namespaces.", + "tags": [ + "gpu", + "tenants", + "admin" + ], + "timezone": "browser", + "editable": true, + "graphTooltip": 1, + "time": { + "from": "now-24h", + "to": "now" + }, + "fiscalYearStartMonth": 0, + "schemaVersion": 42, + "panels": [ + { + "type": "row", + "collapsed": false, + "title": "Tenant fleet snapshot", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 0 + }, + "id": 1, + "panels": [] + }, + { + "type": "stat", + "id": 2, + "targets": [ + { + "expr": "count(namespace:gpu_count:allocated{namespace=~\"$namespace\"} \u003e 0)", + "refId": "A" + } + ], + "title": "Active tenants", + "description": "Namespaces with at least one GPU request (Pending + Running).", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 4, + "w": 4, + "x": 0, + "y": 1 + }, + "repeatDirection": "h", + "options": { + "graphMode": "none", + "colorMode": "value", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "short", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "blue" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "stat", + "id": 3, + "targets": [ + { + "expr": "sum(namespace:gpu_count:allocated{namespace=~\"$namespace\"})", + "refId": "A" + } + ], + "title": "Allocated GPUs", + "description": "Sum of kube GPU requests across tenants. Billable view — includes Pending pods.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 4, + "w": 4, + "x": 4, + "y": 1 + }, + "repeatDirection": "h", + "options": { + "graphMode": "none", + "colorMode": "background", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "short", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "blue" + }, + { + "value": 1, + "color": "green" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "stat", + "id": 4, + "targets": [ + { + "expr": "cluster:gpu_util:avg", + "refId": "A" + } + ], + "title": "Avg cluster GPU util", + "description": "Average NVML utilization across all GPUs in the cluster. Cluster-wide — DCGM hardware metrics cannot be attributed to workload namespaces.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 4, + "w": 4, + "x": 8, + "y": 1 + }, + "repeatDirection": "h", + "options": { + "graphMode": "area", + "colorMode": "value", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "percent", + "min": 0, + "max": 100, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "blue" + }, + { + "value": 30, + "color": "green" + }, + { + "value": 80, + "color": "orange" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "stat", + "id": 5, + "targets": [ + { + "expr": "cluster:gpu_power_watts:sum", + "refId": "A" + } + ], + "title": "Total GPU power", + "description": "Live power draw summed across all GPUs in the cluster. Cluster-wide — DCGM hardware metrics cannot be attributed to workload namespaces.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 4, + "w": 4, + "x": 12, + "y": 1 + }, + "repeatDirection": "h", + "options": { + "graphMode": "area", + "colorMode": "value", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "watt", + "decimals": 0, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "green" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "stat", + "id": 6, + "targets": [ + { + "expr": "sum(sum_over_time(node:power_watts:sum[24h:1m])) / 60 / 1000", + "refId": "A" + } + ], + "title": "Energy (last 24h)", + "description": "Integrated cluster-wide GPU power draw over 24h. Cluster-wide — DCGM hardware metrics cannot be attributed to workload namespaces.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 4, + "w": 4, + "x": 16, + "y": 1 + }, + "repeatDirection": "h", + "options": { + "graphMode": "none", + "colorMode": "value", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "kwatth", + "decimals": 1, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "green" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "stat", + "id": 7, + "targets": [ + { + "expr": "sum(sum_over_time(namespace:gpu_count:allocated{namespace=~\"$namespace\"}[24h:1m])) / 60", + "refId": "A" + } + ], + "title": "GPU-hours (last 24h)", + "description": "Sum of (GPU count × time requested) across all tenants over the last 24h. Integrated from kube requests, so Pending pods count.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 4, + "w": 4, + "x": 20, + "y": 1 + }, + "repeatDirection": "h", + "options": { + "graphMode": "none", + "colorMode": "value", + "justifyMode": "auto", + "textMode": "value", + "wideLayout": true, + "showPercentChange": false, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "percentChangeColorMode": "standard", + "orientation": "" + }, + "fieldConfig": { + "defaults": { + "unit": "short", + "decimals": 1, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "blue" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "row", + "collapsed": false, + "title": "Current allocation", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 5 + }, + "id": 10, + "panels": [] + }, + { + "type": "bargauge", + "id": 11, + "targets": [ + { + "expr": "namespace:gpu_count:allocated{namespace=~\"$namespace\"}", + "legendFormat": "{{namespace}}", + "refId": "A" + } + ], + "title": "GPUs per namespace", + "description": "Kube GPU requests per tenant (billable). Includes Pending pods.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 0, + "y": 6 + }, + "repeatDirection": "h", + "options": { + "displayMode": "gradient", + "valueMode": "color", + "namePlacement": "auto", + "showUnfilled": true, + "sizing": "auto", + "minVizWidth": 8, + "minVizHeight": 16, + "legend": { + "displayMode": "list", + "placement": "bottom", + "showLegend": false, + "calcs": [] + }, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "maxVizHeight": 300, + "orientation": "horizontal" + }, + "fieldConfig": { + "defaults": { + "unit": "short", + "min": 0, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "green" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "bargauge", + "id": 12, + "targets": [ + { + "expr": "node:fb_used_bytes:sum", + "legendFormat": "{{Hostname}}", + "refId": "A" + } + ], + "title": "VRAM used per node", + "description": "FB memory actively used per node. DCGM hardware metrics cannot be attributed to workload namespaces — shown per-node instead.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 12, + "y": 6 + }, + "repeatDirection": "h", + "options": { + "displayMode": "gradient", + "valueMode": "color", + "namePlacement": "auto", + "showUnfilled": true, + "sizing": "auto", + "minVizWidth": 8, + "minVizHeight": 16, + "legend": { + "displayMode": "list", + "placement": "bottom", + "showLegend": false, + "calcs": [] + }, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "maxVizHeight": 300, + "orientation": "horizontal" + }, + "fieldConfig": { + "defaults": { + "unit": "bytes", + "min": 0, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "green" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "row", + "collapsed": false, + "title": "Utilization (per-node)", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 14 + }, + "id": 20, + "panels": [] + }, + { + "type": "timeseries", + "id": 21, + "targets": [ + { + "expr": "node:gpu_util:avg", + "legendFormat": "{{Hostname}}", + "refId": "A" + } + ], + "title": "NVML utilization per node", + "description": "Per-node NVML utilization. Stacked to show fleet-wide pressure over time. DCGM hardware metrics cannot be attributed to workload namespaces — shown per-node instead.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 0, + "y": 15 + }, + "repeatDirection": "h", + "options": { + "legend": { + "displayMode": "table", + "placement": "bottom", + "showLegend": false, + "calcs": [ + "mean", + "max" + ] + }, + "tooltip": { + "mode": "multi", + "sort": "" + } + }, + "fieldConfig": { + "defaults": { + "unit": "percent", + "min": 0, + "custom": { + "drawStyle": "line", + "lineInterpolation": "linear", + "fillOpacity": 40, + "showPoints": "never", + "stacking": { + "mode": "normal" + } + } + }, + "overrides": [] + } + }, + { + "type": "timeseries", + "id": 22, + "targets": [ + { + "expr": "node:tensor_active:avg * 100", + "legendFormat": "{{Hostname}}", + "refId": "A" + } + ], + "title": "Tensor saturation per node", + "description": "Real tensor core load per node. Useful to spot nodes with idle tensor cores. DCGM hardware metrics cannot be attributed to workload namespaces — shown per-node instead.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 12, + "y": 15 + }, + "repeatDirection": "h", + "options": { + "legend": { + "displayMode": "table", + "placement": "bottom", + "showLegend": false, + "calcs": [ + "mean", + "max" + ] + }, + "tooltip": { + "mode": "multi", + "sort": "" + } + }, + "fieldConfig": { + "defaults": { + "unit": "percent", + "min": 0, + "custom": { + "drawStyle": "line", + "lineInterpolation": "linear", + "fillOpacity": 40, + "showPoints": "never", + "stacking": { + "mode": "normal" + } + } + }, + "overrides": [] + } + }, + { + "type": "row", + "collapsed": false, + "title": "Power \u0026 allocation trend", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 23 + }, + "id": 30, + "panels": [] + }, + { + "type": "timeseries", + "id": 31, + "targets": [ + { + "expr": "node:power_watts:sum", + "legendFormat": "{{Hostname}}", + "refId": "A" + } + ], + "title": "Power draw per node", + "description": "Per-node GPU power draw. Stacked to see cluster-wide energy burn over time. DCGM hardware metrics cannot be attributed to workload namespaces — shown per-node instead.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 0, + "y": 24 + }, + "repeatDirection": "h", + "options": { + "legend": { + "displayMode": "table", + "placement": "bottom", + "showLegend": false, + "calcs": [ + "mean", + "max" + ] + }, + "tooltip": { + "mode": "multi", + "sort": "" + } + }, + "fieldConfig": { + "defaults": { + "unit": "watt", + "min": 0, + "custom": { + "drawStyle": "line", + "lineInterpolation": "linear", + "fillOpacity": 40, + "showPoints": "never", + "stacking": { + "mode": "normal" + } + } + }, + "overrides": [] + } + }, + { + "type": "timeseries", + "id": 32, + "targets": [ + { + "expr": "namespace:gpu_count:allocated{namespace=~\"$namespace\"}", + "legendFormat": "{{namespace}}", + "refId": "A" + } + ], + "title": "Allocated GPUs per namespace", + "description": "Kube-requested GPU count per tenant over time. Step-after interpolation reflects discrete allocation changes.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 8, + "w": 12, + "x": 12, + "y": 24 + }, + "repeatDirection": "h", + "options": { + "legend": { + "displayMode": "table", + "placement": "bottom", + "showLegend": false, + "calcs": [ + "mean", + "max" + ] + }, + "tooltip": { + "mode": "multi", + "sort": "" + } + }, + "fieldConfig": { + "defaults": { + "unit": "short", + "min": 0, + "custom": { + "drawStyle": "line", + "lineInterpolation": "stepAfter", + "fillOpacity": 40, + "showPoints": "never", + "stacking": { + "mode": "normal" + } + } + }, + "overrides": [] + } + }, + { + "type": "row", + "collapsed": false, + "title": "Top consumers (live)", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 32 + }, + "id": 40, + "panels": [] + }, + { + "type": "table", + "id": 41, + "targets": [ + { + "expr": "namespace:gpu_count:allocated{namespace=~\"$namespace\"}", + "instant": true, + "range": false, + "format": "table", + "refId": "gpus" + } + ], + "title": "Top consumers (allocation)", + "description": "Live ranking of tenants by GPU allocation. Hardware metrics (util, tensor, power, VRAM) cannot be attributed to workload namespaces — see Fleet and Performance dashboards for per-node/per-GPU hardware views.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 10, + "w": 24, + "x": 0, + "y": 33 + }, + "repeatDirection": "h", + "transformations": [ + { + "id": "organize", + "options": { + "excludeByName": { + "Time": true, + "Time 1": true, + "Time 2": true, + "Time 3": true, + "Time 4": true, + "Time 5": true, + "__name__": true, + "__name__ 1": true, + "__name__ 2": true, + "__name__ 3": true, + "__name__ 4": true, + "__name__ 5": true, + "cluster": true, + "cluster 1": true, + "cluster 2": true, + "cluster 3": true, + "cluster 4": true, + "cluster 5": true, + "endpoint": true, + "endpoint 1": true, + "endpoint 2": true, + "endpoint 3": true, + "endpoint 4": true, + "endpoint 5": true, + "instance": true, + "instance 1": true, + "instance 2": true, + "instance 3": true, + "instance 4": true, + "instance 5": true, + "job": true, + "job 1": true, + "job 2": true, + "job 3": true, + "job 4": true, + "job 5": true, + "prometheus": true, + "prometheus 1": true, + "prometheus 2": true, + "prometheus 3": true, + "prometheus 4": true, + "prometheus 5": true, + "service": true, + "service 1": true, + "service 2": true, + "service 3": true, + "service 4": true, + "service 5": true, + "tenant": true, + "tenant 1": true, + "tenant 2": true, + "tenant 3": true, + "tenant 4": true, + "tenant 5": true, + "tier": true, + "tier 1": true, + "tier 2": true, + "tier 3": true, + "tier 4": true, + "tier 5": true, + "uid": true, + "uid 1": true, + "uid 2": true, + "uid 3": true, + "uid 4": true, + "uid 5": true, + "unit": true, + "unit 1": true, + "unit 2": true, + "unit 3": true, + "unit 4": true, + "unit 5": true + }, + "indexByName": { + "Value #gpus": 1, + "namespace": 0 + }, + "renameByName": { + "Value #gpus": "GPUs", + "namespace": "Namespace" + } + } + } + ], + "options": { + "frameIndex": 0, + "showHeader": true, + "showTypeIcons": false, + "sortBy": [ + { + "displayName": "GPUs", + "desc": true + } + ], + "footer": { + "show": false, + "reducer": [] + }, + "cellHeight": "sm" + } + }, + { + "type": "row", + "collapsed": false, + "title": "Historical usage (last 24h)", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 43 + }, + "id": 50, + "panels": [] + }, + { + "type": "bargauge", + "id": 51, + "targets": [ + { + "expr": "sum_over_time(namespace:gpu_count:allocated{namespace=~\"$namespace\"}[24h:1m]) / 60", + "legendFormat": "{{namespace}}", + "refId": "A" + } + ], + "title": "GPU-hours per namespace (last 24h)", + "description": "Billable GPU-hours per tenant over the last 24h window.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 7, + "w": 12, + "x": 0, + "y": 44 + }, + "repeatDirection": "h", + "options": { + "displayMode": "gradient", + "valueMode": "color", + "namePlacement": "auto", + "showUnfilled": true, + "sizing": "auto", + "minVizWidth": 8, + "minVizHeight": 16, + "legend": { + "displayMode": "list", + "placement": "bottom", + "showLegend": false, + "calcs": [] + }, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "maxVizHeight": 300, + "orientation": "horizontal" + }, + "fieldConfig": { + "defaults": { + "unit": "short", + "decimals": 1, + "min": 0, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "blue" + } + ] + } + }, + "overrides": [] + } + }, + { + "type": "bargauge", + "id": 52, + "targets": [ + { + "expr": "sum_over_time(node:power_watts:sum[24h:1m]) / 60 / 1000", + "legendFormat": "{{Hostname}}", + "refId": "A" + } + ], + "title": "Energy per node (last 24h)", + "description": "kWh consumed per node over the last 24h. Integrated from 1-minute power samples. DCGM hardware metrics cannot be attributed to workload namespaces — shown per-node instead.", + "transparent": false, + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "gridPos": { + "h": 7, + "w": 12, + "x": 12, + "y": 44 + }, + "repeatDirection": "h", + "options": { + "displayMode": "gradient", + "valueMode": "color", + "namePlacement": "auto", + "showUnfilled": true, + "sizing": "auto", + "minVizWidth": 8, + "minVizHeight": 16, + "legend": { + "displayMode": "list", + "placement": "bottom", + "showLegend": false, + "calcs": [] + }, + "reduceOptions": { + "calcs": [ + "lastNotNull" + ] + }, + "maxVizHeight": 300, + "orientation": "horizontal" + }, + "fieldConfig": { + "defaults": { + "unit": "kwatth", + "decimals": 2, + "min": 0, + "thresholds": { + "mode": "absolute", + "steps": [ + { + "value": null, + "color": "green" + } + ] + } + }, + "overrides": [] + } + } + ], + "templating": { + "list": [ + { + "type": "datasource", + "name": "ds_prometheus", + "label": "Prometheus", + "skipUrlSync": false, + "query": "prometheus", + "current": { + "selected": false, + "text": "default", + "value": "default" + }, + "multi": false, + "allowCustomValue": true, + "includeAll": false, + "regex": "", + "auto": false, + "auto_min": "10s", + "auto_count": 30 + }, + { + "type": "query", + "name": "namespace", + "label": "Namespace", + "skipUrlSync": false, + "query": "label_values(namespace:gpu_count:allocated, namespace)", + "datasource": { + "type": "prometheus", + "uid": "$ds_prometheus" + }, + "current": { + "selected": true, + "text": "All", + "value": "$__all" + }, + "multi": true, + "allowCustomValue": true, + "refresh": 2, + "sort": 1, + "includeAll": true, + "auto": false, + "auto_min": "10s", + "auto_count": 30 + } + ] + }, + "annotations": {} +} diff --git a/docs/changelogs/v1.3.1.md b/docs/changelogs/v1.3.1.md new file mode 100644 index 00000000..0f149d0d --- /dev/null +++ b/docs/changelogs/v1.3.1.md @@ -0,0 +1,37 @@ + + +# v1.3.1 (2026-04-28) + +Patch release covering a TenantNamespace IDOR fix in the API, a destructive `post-upgrade` hook removed from the etcd chart, kamaji controller stability, a `linstor-csi` bump that fixes live migration on Protocol-A/B DRBD resources, the missing `linstor-gui` build wiring, and a velero RBAC fix that unblocked installs on bundles without Velero. + +## Security + +* **fix(api): prevent IDOR in TenantNamespace Get and Watch handlers**: Two IDOR (Insecure Direct Object Reference) vulnerabilities allowed authenticated users to read TenantNamespace metadata they had no RoleBinding for. The `Get` and `Watch` handlers now go through a new `hasAccessToNamespace()` helper that lists RoleBindings scoped only to the target namespace (orders of magnitude faster than the previous all-cluster scan), returns `NotFound` instead of leaking existence on unauthorized access, and applies the same check on the `Watch` filter path. Includes regression tests for the unauthorized paths. ([**@IvanHunters**](https://github.com/IvanHunters) in #2471, backport #2524) + +## Features + +* **feat(linstor): bump linstor-csi to v1.10.6 with Protocol-C dual-attach fix**: Live migration of KubeVirt VMs on Protocol-A/B (async) DRBD volumes no longer fails with `Protocol C required`. `linstor-csi` v1.10.6 now installs a `Protocol=C` override on the resource-definition during dual-attach and reverts it on detach, so `replicated-async` StorageClasses and other Protocol-A/B resource groups support live migration without manual `drbdadm adjust` intervention. ([**@kvaps**](https://github.com/kvaps) in #2496, backport #2505) + +## Fixes + +* **fix(backups): move velero-configmap Role to velero chart**: The `backupstrategy-controller` (a default package) declared a Role/RoleBinding scoped to the `cozy-velero` namespace for managing `ResourceModifier` ConfigMaps. On bundles where Velero was not enabled, that namespace did not exist and the HelmRelease failed with `namespaces "cozy-velero" not found`, blocking installation. The Role/RoleBinding now lives in the velero chart, so it is created only when velero is actually deployed. ([**@myasnikovdaniil**](https://github.com/myasnikovdaniil) in #2459, backport #2467) + +* **fix(etcd): remove destructive post-upgrade cert-regeneration hook**: The etcd chart ran a `post-upgrade` Helm hook on every upgrade that deleted etcd TLS Secrets (`etcd-ca-tls`, `etcd-peer-ca-tls`, `etcd-client-tls`, `etcd-peer-tls`, `etcd-server-tls`) and then deleted etcd pods, forcing cert-manager to re-issue the entire etcd CA chain. On clusters with Kamaji-managed tenant control planes this put every tenant `kube-apiserver` into CrashLoopBackOff until each DataStore was manually re-reconciled. The hook was a one-shot `2.6.0 → 2.6.1` migration that became a permanent footgun once chart versioning moved to `0.0.0+` (always `< 2.6.1` per semver) and after the underlying `rotationPolicy: Always` issue was fixed in `47d81f70`. The hook is now removed entirely. ([**@myasnikovdaniil**](https://github.com/myasnikovdaniil) in #2462, backport #2511) + +* **fix(kamaji): increase memory limits and add startup probe**: The kamaji controller frequently entered CrashLoopBackOff due to OOMKills (exit 137) within ~20–25 seconds of startup, with the readiness probe failing while the controller was still finishing initialization. Memory limit raised from 500Mi to 512Mi, request from 100Mi to 256Mi, and a 60-second startup probe (12 attempts × 5s periods) is added so the controller has room to boot before liveness/readiness probes engage. ([**@IvanHunters**](https://github.com/IvanHunters) in #2421, backport #2491) + +## Build + +* **build(linstor): include linstor-gui in root image build target**: The `linstor-gui` package (added in #2382) was never wired into the root `Makefile`'s `build:` target, so CI never built or published the image. `ghcr.io/cozystack/cozystack/linstor-gui` returned `NAME_UNKNOWN` and `values.yaml` stayed pinned to `tag: 2.3.0` without a digest. The missing build line is added so the next CI run publishes the image and the per-package `Makefile` digest-pins `values.yaml` automatically. ([**@myasnikovdaniil**](https://github.com/myasnikovdaniil) in #2498, backport #2518) + +## Contributors + +Thanks to everyone who contributed to this patch release: + +* [**@IvanHunters**](https://github.com/IvanHunters) +* [**@kvaps**](https://github.com/kvaps) +* [**@myasnikovdaniil**](https://github.com/myasnikovdaniil) + +**Full Changelog**: https://github.com/cozystack/cozystack/compare/v1.3.0...v1.3.1 diff --git a/hack/check-gpu-recording-rules.bats b/hack/check-gpu-recording-rules.bats new file mode 100644 index 00000000..bae6785e --- /dev/null +++ b/hack/check-gpu-recording-rules.bats @@ -0,0 +1,206 @@ +#!/usr/bin/env bats +# ----------------------------------------------------------------------------- +# Cross-validation between GPU recording rules, the dashboards that consume +# them, and the DCGM Exporter metric set the cluster actually scrapes. Catches: +# +# 1. dangling references — a dashboard query mentions a recording rule name +# that doesn't exist in gpu-recording.rules.yaml. This is the bug the +# pre-merge review caught: gpu-efficiency.json shipped panels keyed on +# pod:tensor_saturation:avg5m without the rule being defined, so the +# panel showed "No data" everywhere. +# +# 2. typos in rule names — same bug class, manifested as a single-character +# difference between rule and reference. +# +# 3. undeclared DCGM metrics — a dashboard query or recording rule mentions +# a DCGM_FI_* metric that is neither in the upstream default CSV nor in +# the project's custom CSV (dcgm-custom-metrics.yaml), meaning DCGM +# Exporter would never emit it and the panel silently shows "No data". +# Example regression: gpu-fleet.json shipped a TDP panel referencing +# DCGM_FI_DEV_POWER_MGMT_LIMIT before the custom CSV declared it. +# +# Scope: only dashboards listed in packages/system/monitoring/dashboards-infra.list +# under the "gpu/" prefix (i.e. shipped to production), not every JSON file in +# dashboards/gpu/. Untracked drafts stay out of scope on purpose — adding one +# to dashboards-infra.list is what brings it under the test. +# +# Reverse direction (rule defined but never consumed) is intentionally NOT +# enforced: some rules exist for ad-hoc PromQL or upcoming dashboards. Treat +# unused rules as an editorial concern, not a regression. +# +# Title syntax constraints from cozytest.sh's awk parser: +# - Titles delimited by ASCII double quotes; embedded quotes truncate. +# - Only [A-Za-z0-9] from the title survives into the function name; titles +# differing only in punctuation collapse to the same function. +# +# Run with: hack/cozytest.sh hack/check-gpu-recording-rules.bats +# ----------------------------------------------------------------------------- + +REPO_ROOT="$(cd "$(dirname "${BATS_TEST_FILENAME:-$0}")/.." && pwd)" +RULES_FILE="$REPO_ROOT/packages/system/monitoring-agents/alerts/gpu-recording.rules.yaml" +DASHBOARDS_LIST="$REPO_ROOT/packages/system/monitoring/dashboards-infra.list" +DASHBOARDS_DIR="$REPO_ROOT/dashboards" +DCGM_DEFAULT_CSV="$REPO_ROOT/hack/dcgm-default-counters.csv" +DCGM_CUSTOM_CSV="$REPO_ROOT/packages/system/gpu-operator/examples/dcgm-custom-metrics.yaml" + +# Extract the set of "- record: NAME" entries from the rules YAML. +# Outputs one rule name per line, sorted and deduplicated. +extract_rules() { + awk '/^[[:space:]]*-[[:space:]]*record:[[:space:]]/ { + sub(/^[[:space:]]*-[[:space:]]*record:[[:space:]]*/, "") + sub(/[[:space:]]*$/, "") + print + }' "$RULES_FILE" | sort -u +} + +# Extract the set of recording-rule references from a dashboard JSON. +# A recording-rule reference is matched by the pattern +# :(:)+ +# where each is [a-z0-9_]. Raw DCGM metrics (DCGM_FI_*), +# kube-state-metrics (kube_*) and similar uppercase / single-word metric +# names do not match because the leading segment must be lowercase and the +# whole expression must contain at least two ':' characters. +extract_refs() { + json_file=$1 + # Prometheus convention allows 2-segment rule names (level:metric); this + # regex is tuned to the 3+ segment convention used in this repo + # (level:metric:op — e.g. cluster:gpu_count:total). Update if future + # rules use 2 segments, otherwise they will be silently skipped. + grep -hoE '[a-z][a-z0-9_]*:[a-z0-9_]+:[a-z0-9_]+' "$json_file" | sort -u +} + +# Resolve "gpu/foo" -> "$DASHBOARDS_DIR/gpu/foo.json" +list_tracked_gpu_dashboards() { + awk '/^gpu\// { print $0 ".json" }' "$DASHBOARDS_LIST" +} + +# Extract the set of DCGM_FI_* metric names declared in a CSV file. Handles +# both the upstream-style default CSV (unindented) and the ConfigMap-style +# custom CSV (YAML-indented). A declaration line starts — after any leading +# whitespace — with "DCGM_FI_," ; comment lines begin with "#" and are +# skipped. Uses POSIX awk's match()+RSTART/RLENGTH so no GNU extensions +# are required. +extract_csv_metrics() { + file=$1 + awk ' + { + line = $0 + sub(/^[[:space:]]+/, "", line) + if (line ~ /^#/) next + if (match(line, /^DCGM_FI_[A-Z0-9_]+/)) { + print substr(line, RSTART, RLENGTH) + } + } + ' "$file" | sort -u +} + +# Extract the set of DCGM_FI_* metric references from a text file (dashboard +# JSON or rules YAML). A DCGM metric name has at least two underscore-delimited +# segments after the "DCGM_FI_" prefix (e.g. DCGM_FI_DEV_GPU_UTIL, DCGM_FI_PROF_ +# PIPE_TENSOR_ACTIVE, DCGM_FI_DRIVER_VERSION). Requiring two segments keeps +# the matcher from latching onto glob stubs like "DCGM_FI_DEV_*_VIOLATION" that +# appear in comments. +extract_dcgm_refs() { + file=$1 + grep -hoE 'DCGM_FI_[A-Z0-9]+(_[A-Z0-9]+)+' "$file" | sort -u +} + +@test "every recording rule reference in tracked GPU dashboards has a matching record" { + TMP=$(mktemp -d) + trap 'rm -rf "$TMP"' EXIT + + extract_rules > "$TMP/rules.txt" + [ -s "$TMP/rules.txt" ] || { echo "no recording rules extracted from $RULES_FILE" >&2; exit 1; } + + list_tracked_gpu_dashboards > "$TMP/dashboards.txt" + [ -s "$TMP/dashboards.txt" ] || { echo "no gpu/* dashboards listed in $DASHBOARDS_LIST" >&2; exit 1; } + + failed=0 + while IFS= read -r dashboard_rel; do + dashboard="$DASHBOARDS_DIR/$dashboard_rel" + if [ ! -f "$dashboard" ]; then + echo "ERROR: dashboard listed but file missing: $dashboard" >&2 + failed=1 + continue + fi + + extract_refs "$dashboard" > "$TMP/refs.txt" + # comm -23: lines unique to refs.txt (referenced but not defined) + # Both inputs must be sorted; extract_* helpers already sort. + comm -23 "$TMP/refs.txt" "$TMP/rules.txt" > "$TMP/missing.txt" + if [ -s "$TMP/missing.txt" ]; then + echo "ERROR: $dashboard_rel references undefined recording rules:" >&2 + sed 's/^/ - /' "$TMP/missing.txt" >&2 + failed=1 + fi + done < "$TMP/dashboards.txt" + + [ "$failed" -eq 0 ] +} + +@test "every DCGM metric referenced in tracked dashboards and rules is declared" { + TMP=$(mktemp -d) + trap 'rm -rf "$TMP"' EXIT + + [ -f "$DCGM_DEFAULT_CSV" ] || { echo "missing $DCGM_DEFAULT_CSV" >&2; exit 1; } + [ -f "$DCGM_CUSTOM_CSV" ] || { echo "missing $DCGM_CUSTOM_CSV" >&2; exit 1; } + + { + extract_csv_metrics "$DCGM_DEFAULT_CSV" + extract_csv_metrics "$DCGM_CUSTOM_CSV" + } | sort -u > "$TMP/declared.txt" + [ -s "$TMP/declared.txt" ] || { echo "no DCGM metrics extracted from CSVs" >&2; exit 1; } + + list_tracked_gpu_dashboards > "$TMP/dashboards.txt" + [ -s "$TMP/dashboards.txt" ] || { echo "no gpu/* dashboards listed in $DASHBOARDS_LIST" >&2; exit 1; } + + failed=0 + + # Dashboard coverage — every dashboard's DCGM references must resolve. + while IFS= read -r dashboard_rel; do + dashboard="$DASHBOARDS_DIR/$dashboard_rel" + [ -f "$dashboard" ] || continue # handled by the existence test + extract_dcgm_refs "$dashboard" > "$TMP/refs.txt" + [ -s "$TMP/refs.txt" ] || continue # dashboard relies entirely on recording rules + comm -23 "$TMP/refs.txt" "$TMP/declared.txt" > "$TMP/missing.txt" + if [ -s "$TMP/missing.txt" ]; then + echo "ERROR: $dashboard_rel references DCGM metrics not declared in any CSV:" >&2 + sed 's/^/ - /' "$TMP/missing.txt" >&2 + failed=1 + fi + done < "$TMP/dashboards.txt" + + # Rules coverage — recording rules consume DCGM directly, so their set + # must be declared too, otherwise derived series on every dashboard + # collapse to empty. + extract_dcgm_refs "$RULES_FILE" > "$TMP/rule-refs.txt" + if [ -s "$TMP/rule-refs.txt" ]; then + comm -23 "$TMP/rule-refs.txt" "$TMP/declared.txt" > "$TMP/rule-missing.txt" + if [ -s "$TMP/rule-missing.txt" ]; then + echo "ERROR: gpu-recording.rules.yaml references DCGM metrics not declared in any CSV:" >&2 + sed 's/^/ - /' "$TMP/rule-missing.txt" >&2 + failed=1 + fi + fi + + [ "$failed" -eq 0 ] +} + +@test "every tracked GPU dashboard listed in dashboards-infra.list exists on disk" { + TMP=$(mktemp -d) + trap 'rm -rf "$TMP"' EXIT + + list_tracked_gpu_dashboards > "$TMP/dashboards.txt" + [ -s "$TMP/dashboards.txt" ] || { echo "no gpu/* dashboards listed in $DASHBOARDS_LIST" >&2; exit 1; } + + failed=0 + while IFS= read -r dashboard_rel; do + dashboard="$DASHBOARDS_DIR/$dashboard_rel" + if [ ! -f "$dashboard" ]; then + echo "ERROR: $dashboard_rel listed in dashboards-infra.list but $dashboard does not exist" >&2 + failed=1 + fi + done < "$TMP/dashboards.txt" + + [ "$failed" -eq 0 ] +} diff --git a/hack/dcgm-default-counters.csv b/hack/dcgm-default-counters.csv new file mode 100644 index 00000000..b5e94540 --- /dev/null +++ b/hack/dcgm-default-counters.csv @@ -0,0 +1,104 @@ +# Snapshot of the upstream DCGM Exporter default-counters.csv used as a +# fixture by hack/check-gpu-recording-rules.bats. The test verifies that +# every DCGM_FI_* metric referenced by a tracked GPU dashboard is either +# declared here (upstream defaults) or in +# packages/system/gpu-operator/examples/dcgm-custom-metrics.yaml +# (the project's custom CSV). +# +# Source: https://github.com/NVIDIA/dcgm-exporter/blob/4.1.1-4.0.4/etc/default-counters.csv +# Pinned to the DCGM Exporter image tag shipped by the gpu-operator +# chart under packages/system/gpu-operator/charts/gpu-operator/values.yaml +# (dcgmExporter.version = 4.1.1-4.0.4-ubuntu22.04). When that image is +# bumped, refresh this file from the matching tag in the NVIDIA/dcgm-exporter +# repo and re-run `./hack/cozytest.sh hack/check-gpu-recording-rules.bats`. + +# Format +# If line starts with a '#' it is considered a comment +# DCGM FIELD, Prometheus metric type, help message + +# Clocks +DCGM_FI_DEV_SM_CLOCK, gauge, SM clock frequency (in MHz). +DCGM_FI_DEV_MEM_CLOCK, gauge, Memory clock frequency (in MHz). + +# Temperature +DCGM_FI_DEV_MEMORY_TEMP, gauge, Memory temperature (in C). +DCGM_FI_DEV_GPU_TEMP, gauge, GPU temperature (in C). + +# Power +DCGM_FI_DEV_POWER_USAGE, gauge, Power draw (in W). +DCGM_FI_DEV_TOTAL_ENERGY_CONSUMPTION, counter, Total energy consumption since boot (in mJ). + +# PCIE +# DCGM_FI_PROF_PCIE_TX_BYTES, counter, Total number of bytes transmitted through PCIe TX via NVML. +# DCGM_FI_PROF_PCIE_RX_BYTES, counter, Total number of bytes received through PCIe RX via NVML. +DCGM_FI_DEV_PCIE_REPLAY_COUNTER, counter, Total number of PCIe retries. + +# Utilization (the sample period varies depending on the product) +DCGM_FI_DEV_GPU_UTIL, gauge, GPU utilization (in %). +DCGM_FI_DEV_MEM_COPY_UTIL, gauge, Memory utilization (in %). +DCGM_FI_DEV_ENC_UTIL, gauge, Encoder utilization (in %). +DCGM_FI_DEV_DEC_UTIL , gauge, Decoder utilization (in %). + +# Errors and violations +DCGM_FI_DEV_XID_ERRORS, gauge, Value of the last XID error encountered. +# DCGM_FI_DEV_POWER_VIOLATION, counter, Throttling duration due to power constraints (in us). +# DCGM_FI_DEV_THERMAL_VIOLATION, counter, Throttling duration due to thermal constraints (in us). +# DCGM_FI_DEV_SYNC_BOOST_VIOLATION, counter, Throttling duration due to sync-boost constraints (in us). +# DCGM_FI_DEV_BOARD_LIMIT_VIOLATION, counter, Throttling duration due to board limit constraints (in us). +# DCGM_FI_DEV_LOW_UTIL_VIOLATION, counter, Throttling duration due to low utilization (in us). +# DCGM_FI_DEV_RELIABILITY_VIOLATION, counter, Throttling duration due to reliability constraints (in us). + +# Memory usage +DCGM_FI_DEV_FB_FREE, gauge, Framebuffer memory free (in MiB). +DCGM_FI_DEV_FB_USED, gauge, Framebuffer memory used (in MiB). + +# ECC +# DCGM_FI_DEV_ECC_SBE_VOL_TOTAL, counter, Total number of single-bit volatile ECC errors. +# DCGM_FI_DEV_ECC_DBE_VOL_TOTAL, counter, Total number of double-bit volatile ECC errors. +# DCGM_FI_DEV_ECC_SBE_AGG_TOTAL, counter, Total number of single-bit persistent ECC errors. +# DCGM_FI_DEV_ECC_DBE_AGG_TOTAL, counter, Total number of double-bit persistent ECC errors. + +# Retired pages +# DCGM_FI_DEV_RETIRED_SBE, counter, Total number of retired pages due to single-bit errors. +# DCGM_FI_DEV_RETIRED_DBE, counter, Total number of retired pages due to double-bit errors. +# DCGM_FI_DEV_RETIRED_PENDING, counter, Total number of pages pending retirement. + +# NVLink +# DCGM_FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT_TOTAL, counter, Total number of NVLink flow-control CRC errors. +# DCGM_FI_DEV_NVLINK_CRC_DATA_ERROR_COUNT_TOTAL, counter, Total number of NVLink data CRC errors. +# DCGM_FI_DEV_NVLINK_REPLAY_ERROR_COUNT_TOTAL, counter, Total number of NVLink retries. +# DCGM_FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_TOTAL, counter, Total number of NVLink recovery errors. +DCGM_FI_DEV_NVLINK_BANDWIDTH_TOTAL, counter, Total number of NVLink bandwidth counters for all lanes. +# DCGM_FI_DEV_NVLINK_BANDWIDTH_L0, counter, The number of bytes of active NVLink rx or tx data including both header and payload. + +# VGPU License status +DCGM_FI_DEV_VGPU_LICENSE_STATUS, gauge, vGPU License status + +# Remapped rows +DCGM_FI_DEV_UNCORRECTABLE_REMAPPED_ROWS, counter, Number of remapped rows for uncorrectable errors +DCGM_FI_DEV_CORRECTABLE_REMAPPED_ROWS, counter, Number of remapped rows for correctable errors +DCGM_FI_DEV_ROW_REMAP_FAILURE, gauge, Whether remapping of rows has failed + +# Static configuration information. These appear as labels on the other metrics +DCGM_FI_DRIVER_VERSION, label, Driver Version +# DCGM_FI_NVML_VERSION, label, NVML Version +# DCGM_FI_DEV_BRAND, label, Device Brand +# DCGM_FI_DEV_SERIAL, label, Device Serial Number +# DCGM_FI_DEV_OEM_INFOROM_VER, label, OEM inforom version +# DCGM_FI_DEV_ECC_INFOROM_VER, label, ECC inforom version +# DCGM_FI_DEV_POWER_INFOROM_VER, label, Power management object inforom version +# DCGM_FI_DEV_INFOROM_IMAGE_VER, label, Inforom image version +# DCGM_FI_DEV_VBIOS_VERSION, label, VBIOS version of the device + +# Datacenter Profiling (DCP) metrics +# NOTE: supported on Nvidia datacenter Volta GPUs and newer +DCGM_FI_PROF_GR_ENGINE_ACTIVE, gauge, Ratio of time the graphics engine is active. +# DCGM_FI_PROF_SM_ACTIVE, gauge, The ratio of cycles an SM has at least 1 warp assigned. +# DCGM_FI_PROF_SM_OCCUPANCY, gauge, The ratio of number of warps resident on an SM. +DCGM_FI_PROF_PIPE_TENSOR_ACTIVE, gauge, Ratio of cycles the tensor (HMMA) pipe is active. +DCGM_FI_PROF_DRAM_ACTIVE, gauge, Ratio of cycles the device memory interface is active sending or receiving data. +# DCGM_FI_PROF_PIPE_FP64_ACTIVE, gauge, Ratio of cycles the fp64 pipes are active. +# DCGM_FI_PROF_PIPE_FP32_ACTIVE, gauge, Ratio of cycles the fp32 pipes are active. +# DCGM_FI_PROF_PIPE_FP16_ACTIVE, gauge, Ratio of cycles the fp16 pipes are active. +DCGM_FI_PROF_PCIE_TX_BYTES, gauge, The rate of data transmitted over the PCIe bus - including both protocol headers and data payloads - in bytes per second. +DCGM_FI_PROF_PCIE_RX_BYTES, gauge, The rate of data received over the PCIe bus - including both protocol headers and data payloads - in bytes per second. diff --git a/packages/apps/kubernetes/README.md b/packages/apps/kubernetes/README.md index ecbbc31b..bed1117a 100644 --- a/packages/apps/kubernetes/README.md +++ b/packages/apps/kubernetes/README.md @@ -128,6 +128,9 @@ See the reference for components utilized in this service: | `addons.gpuOperator` | NVIDIA GPU Operator. | `object` | `{}` | | `addons.gpuOperator.enabled` | Enable GPU Operator. | `bool` | `false` | | `addons.gpuOperator.valuesOverride` | Custom Helm values overrides. | `object` | `{}` | +| `addons.hami` | HAMi GPU virtualization middleware. | `object` | `{}` | +| `addons.hami.enabled` | Enable HAMi (requires GPU Operator). | `bool` | `false` | +| `addons.hami.valuesOverride` | Custom Helm values overrides. | `object` | `{}` | | `addons.fluxcd` | FluxCD GitOps operator. | `object` | `{}` | | `addons.fluxcd.enabled` | Enable FluxCD. | `bool` | `false` | | `addons.fluxcd.valuesOverride` | Custom Helm values overrides. | `object` | `{}` | diff --git a/packages/apps/kubernetes/templates/helmreleases/gpu-operator.yaml b/packages/apps/kubernetes/templates/helmreleases/gpu-operator.yaml index e243bfad..655dc868 100644 --- a/packages/apps/kubernetes/templates/helmreleases/gpu-operator.yaml +++ b/packages/apps/kubernetes/templates/helmreleases/gpu-operator.yaml @@ -1,3 +1,11 @@ +{{- define "cozystack.defaultGpuOperatorValues" -}} +{{- if .Values.addons.hami.enabled }} +gpu-operator: + devicePlugin: + enabled: false +{{- end }} +{{- end }} + {{- if and .Values.addons.gpuOperator.enabled .Values._namespace.etcd }} apiVersion: helm.toolkit.fluxcd.io/v2 kind: HelmRelease @@ -29,9 +37,12 @@ spec: force: true remediation: retries: -1 - {{- with .Values.addons.gpuOperator.valuesOverride }} + {{- $defaults := fromYaml (include "cozystack.defaultGpuOperatorValues" .) }} + {{- $overrides := deepCopy (default (dict) .Values.addons.gpuOperator.valuesOverride) }} + {{- $merged := mergeOverwrite (default (dict) $defaults) $overrides }} + {{- if $merged }} values: - {{- toYaml . | nindent 4 }} + {{- toYaml $merged | nindent 4 }} {{- end }} dependsOn: diff --git a/packages/apps/kubernetes/templates/helmreleases/hami.yaml b/packages/apps/kubernetes/templates/helmreleases/hami.yaml new file mode 100644 index 00000000..f1538c7c --- /dev/null +++ b/packages/apps/kubernetes/templates/helmreleases/hami.yaml @@ -0,0 +1,49 @@ +{{- if and .Values.addons.hami.enabled .Values._namespace.etcd }} +{{- if not .Values.addons.gpuOperator.enabled }} +{{- fail "addons.hami requires addons.gpuOperator to be enabled" }} +{{- end }} +apiVersion: helm.toolkit.fluxcd.io/v2 +kind: HelmRelease +metadata: + name: {{ .Release.Name }}-hami + labels: + cozystack.io/repository: system + cozystack.io/target-cluster-name: {{ .Release.Name }} + sharding.fluxcd.io/key: tenants +spec: + releaseName: hami + chartRef: + kind: ExternalArtifact + name: cozystack-kubernetes-application-kubevirt-kubernetes-hami + namespace: cozy-system + kubeConfig: + secretRef: + name: {{ .Release.Name }}-admin-kubeconfig + key: super-admin.svc + targetNamespace: cozy-hami + storageNamespace: cozy-hami + interval: 5m + timeout: 10m + install: + createNamespace: true + remediation: + retries: -1 + upgrade: + force: true + remediation: + retries: -1 + {{- with .Values.addons.hami.valuesOverride }} + values: + {{- toYaml . | nindent 4 }} + {{- end }} + + dependsOn: + {{- if lookup "helm.toolkit.fluxcd.io/v2" "HelmRelease" .Release.Namespace .Release.Name }} + - name: {{ .Release.Name }} + namespace: {{ .Release.Namespace }} + {{- end }} + - name: {{ .Release.Name }}-cilium + namespace: {{ .Release.Namespace }} + - name: {{ .Release.Name }}-gpu-operator + namespace: {{ .Release.Namespace }} +{{- end }} diff --git a/packages/apps/kubernetes/tests/gpu_operator_hami_test.yaml b/packages/apps/kubernetes/tests/gpu_operator_hami_test.yaml new file mode 100644 index 00000000..44528470 --- /dev/null +++ b/packages/apps/kubernetes/tests/gpu_operator_hami_test.yaml @@ -0,0 +1,99 @@ +suite: GPU Operator HelmRelease HAMi integration tests +templates: + - templates/helmreleases/gpu-operator.yaml +values: + - values-ci.yaml +tests: + - it: should disable devicePlugin when hami is enabled + set: + addons: + gpuOperator: + enabled: true + valuesOverride: {} + hami: + enabled: true + valuesOverride: {} + asserts: + - equal: + path: spec.values.gpu-operator.devicePlugin.enabled + value: false + + - it: should not have values when hami is disabled and no overrides + set: + addons: + gpuOperator: + enabled: true + valuesOverride: {} + hami: + enabled: false + valuesOverride: {} + asserts: + - notExists: + path: spec.values + + - it: should apply hami defaults when valuesOverride key is omitted + set: + addons: + gpuOperator: + enabled: true + hami: + enabled: true + asserts: + - equal: + path: spec.values.gpu-operator.devicePlugin.enabled + value: false + + - it: should allow user overrides to merge with hami defaults + set: + addons: + gpuOperator: + enabled: true + valuesOverride: + gpu-operator: + driver: + enabled: false + hami: + enabled: true + valuesOverride: {} + asserts: + - equal: + path: spec.values.gpu-operator.devicePlugin.enabled + value: false + - equal: + path: spec.values.gpu-operator.driver.enabled + value: false + + - it: should let user explicitly override devicePlugin.enabled to true with hami enabled + set: + addons: + gpuOperator: + enabled: true + valuesOverride: + gpu-operator: + devicePlugin: + enabled: true + driver: + enabled: false + hami: + enabled: true + valuesOverride: {} + asserts: + - equal: + path: spec.values.gpu-operator.devicePlugin.enabled + value: true + - equal: + path: spec.values.gpu-operator.driver.enabled + value: false + + - it: should not render when gpuOperator is disabled + set: + addons: + gpuOperator: + enabled: false + valuesOverride: {} + hami: + enabled: false + valuesOverride: {} + asserts: + - hasDocuments: + count: 0 diff --git a/packages/apps/kubernetes/tests/hami_test.yaml b/packages/apps/kubernetes/tests/hami_test.yaml new file mode 100644 index 00000000..27f7c75b --- /dev/null +++ b/packages/apps/kubernetes/tests/hami_test.yaml @@ -0,0 +1,153 @@ +suite: HAMi HelmRelease tests +templates: + - templates/helmreleases/hami.yaml +values: + - values-ci.yaml +tests: + - it: should not render when hami is disabled + set: + addons: + hami: + enabled: false + valuesOverride: {} + gpuOperator: + enabled: true + valuesOverride: {} + asserts: + - hasDocuments: + count: 0 + + - it: should render HelmRelease when hami is enabled + set: + addons: + hami: + enabled: true + valuesOverride: {} + gpuOperator: + enabled: true + valuesOverride: {} + asserts: + - hasDocuments: + count: 1 + - isKind: + of: HelmRelease + + - it: should fail when gpuOperator is not enabled + set: + addons: + hami: + enabled: true + valuesOverride: {} + gpuOperator: + enabled: false + valuesOverride: {} + asserts: + - failedTemplate: + errorMessage: "addons.hami requires addons.gpuOperator to be enabled" + + - it: should have correct metadata labels + set: + addons: + hami: + enabled: true + valuesOverride: {} + gpuOperator: + enabled: true + valuesOverride: {} + asserts: + - equal: + path: metadata.labels["cozystack.io/repository"] + value: system + - equal: + path: metadata.labels["sharding.fluxcd.io/key"] + value: tenants + + - it: should use ExternalArtifact chartRef + set: + addons: + hami: + enabled: true + valuesOverride: {} + gpuOperator: + enabled: true + valuesOverride: {} + asserts: + - equal: + path: spec.chartRef.kind + value: ExternalArtifact + - equal: + path: spec.chartRef.namespace + value: cozy-system + + - it: should target cozy-hami namespace + set: + addons: + hami: + enabled: true + valuesOverride: {} + gpuOperator: + enabled: true + valuesOverride: {} + asserts: + - equal: + path: spec.targetNamespace + value: cozy-hami + - equal: + path: spec.storageNamespace + value: cozy-hami + + - it: should depend on gpu-operator and cilium + release: + name: test + namespace: test-ns + set: + addons: + hami: + enabled: true + valuesOverride: {} + gpuOperator: + enabled: true + valuesOverride: {} + asserts: + - contains: + path: spec.dependsOn + content: + name: test-cilium + namespace: test-ns + - contains: + path: spec.dependsOn + content: + name: test-gpu-operator + namespace: test-ns + + - it: should not render spec.values when valuesOverride is empty + set: + addons: + hami: + enabled: true + valuesOverride: {} + gpuOperator: + enabled: true + valuesOverride: {} + asserts: + - hasDocuments: + count: 1 + - notExists: + path: spec.values + + - it: should pass through valuesOverride + set: + addons: + hami: + enabled: true + valuesOverride: + hami: + devicePlugin: + deviceSplitCount: 5 + gpuOperator: + enabled: true + valuesOverride: {} + asserts: + - equal: + path: spec.values.hami.devicePlugin.deviceSplitCount + value: 5 diff --git a/packages/apps/kubernetes/values.schema.json b/packages/apps/kubernetes/values.schema.json index 81e9e2b8..a43deaec 100644 --- a/packages/apps/kubernetes/values.schema.json +++ b/packages/apps/kubernetes/values.schema.json @@ -149,6 +149,7 @@ "fluxcd", "gatewayAPI", "gpuOperator", + "hami", "ingressNginx", "monitoringAgents", "velero", @@ -268,6 +269,28 @@ } } }, + "hami": { + "description": "HAMi GPU virtualization middleware.", + "type": "object", + "default": {}, + "required": [ + "enabled", + "valuesOverride" + ], + "properties": { + "enabled": { + "description": "Enable HAMi (requires GPU Operator).", + "type": "boolean", + "default": false + }, + "valuesOverride": { + "description": "Custom Helm values overrides.", + "type": "object", + "default": {}, + "x-kubernetes-preserve-unknown-fields": true + } + } + }, "ingressNginx": { "description": "Ingress-NGINX controller.", "type": "object", diff --git a/packages/apps/kubernetes/values.yaml b/packages/apps/kubernetes/values.yaml index 68ba07b0..d609476c 100644 --- a/packages/apps/kubernetes/values.yaml +++ b/packages/apps/kubernetes/values.yaml @@ -94,6 +94,10 @@ host: "" ## @field {bool} enabled - Enable FluxCD. ## @field {object} valuesOverride - Custom Helm values overrides. +## @typedef {struct} HAMiAddon - HAMi GPU virtualization middleware. +## @field {bool} enabled - Enable HAMi (requires GPU Operator). +## @field {object} valuesOverride - Custom Helm values overrides. + ## @typedef {struct} MonitoringAgentsAddon - Monitoring agents (Fluent Bit, VMAgents). ## @field {bool} enabled - Enable monitoring agents. ## @field {object} valuesOverride - Custom Helm values overrides. @@ -114,6 +118,7 @@ host: "" ## @field {GatewayAPIAddon} gatewayAPI - Gateway API addon. ## @field {IngressNginxAddon} ingressNginx - Ingress-NGINX controller. ## @field {GPUOperatorAddon} gpuOperator - NVIDIA GPU Operator. +## @field {HAMiAddon} hami - HAMi GPU virtualization middleware. ## @field {FluxCDAddon} fluxcd - FluxCD GitOps operator. ## @field {MonitoringAgentsAddon} monitoringAgents - Monitoring agents. ## @field {VerticalPodAutoscalerAddon} verticalPodAutoscaler - Vertical Pod Autoscaler. @@ -137,6 +142,9 @@ addons: gpuOperator: enabled: false valuesOverride: {} + hami: + enabled: false + valuesOverride: {} fluxcd: enabled: false valuesOverride: {} diff --git a/packages/apps/postgres/README.md b/packages/apps/postgres/README.md index 5a550f7a..4cda7284 100644 --- a/packages/apps/postgres/README.md +++ b/packages/apps/postgres/README.md @@ -133,12 +133,13 @@ See: ### Bootstrap (recovery) parameters -| Name | Description | Type | Value | -| ------------------------ | ------------------------------------------------------------------- | -------- | ------- | -| `bootstrap` | Bootstrap configuration. | `object` | `{}` | -| `bootstrap.enabled` | Whether to restore from a backup. | `bool` | `false` | -| `bootstrap.recoveryTime` | Timestamp (RFC3339) for point-in-time recovery; empty means latest. | `string` | `""` | -| `bootstrap.oldName` | Previous cluster name before deletion. | `string` | `""` | +| Name | Description | Type | Value | +| ------------------------ | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -------- | ------- | +| `bootstrap` | Bootstrap configuration. | `object` | `{}` | +| `bootstrap.enabled` | Whether to restore from a backup. | `bool` | `false` | +| `bootstrap.recoveryTime` | Timestamp (RFC3339) for point-in-time recovery; empty means latest. | `string` | `""` | +| `bootstrap.oldName` | Previous cluster name before deletion. | `string` | `""` | +| `bootstrap.serverName` | Barman server name (S3 path prefix) used by the original cluster when writing backups. Set this only when the original cluster had an explicit barmanObjectStore.serverName that differed from its Kubernetes resource name. | `string` | `""` | ## Parameter examples and reference diff --git a/packages/apps/postgres/templates/db.yaml b/packages/apps/postgres/templates/db.yaml index 5c40b2c7..2cdaec1d 100644 --- a/packages/apps/postgres/templates/db.yaml +++ b/packages/apps/postgres/templates/db.yaml @@ -32,6 +32,9 @@ spec: - name: {{ .Values.bootstrap.oldName }} barmanObjectStore: destinationPath: {{ .Values.backup.destinationPath }} + {{- if .Values.bootstrap.serverName }} + serverName: {{ .Values.bootstrap.serverName }} + {{- end }} endpointURL: {{ .Values.backup.endpointURL }} s3Credentials: accessKeyId: diff --git a/packages/apps/postgres/values.schema.json b/packages/apps/postgres/values.schema.json index b2a4aeba..98e29822 100644 --- a/packages/apps/postgres/values.schema.json +++ b/packages/apps/postgres/values.schema.json @@ -254,6 +254,11 @@ "description": "Timestamp (RFC3339) for point-in-time recovery; empty means latest.", "type": "string", "default": "" + }, + "serverName": { + "description": "Barman server name (S3 path prefix) used by the original cluster when writing backups. Set this only when the original cluster had an explicit barmanObjectStore.serverName that differed from its Kubernetes resource name.", + "type": "string", + "default": "" } } } diff --git a/packages/apps/postgres/values.yaml b/packages/apps/postgres/values.yaml index b8f07f63..2ceaa9cf 100644 --- a/packages/apps/postgres/values.yaml +++ b/packages/apps/postgres/values.yaml @@ -154,6 +154,7 @@ backup: ## @field {bool} enabled - Whether to restore from a backup. ## @field {string} [recoveryTime] - Timestamp (RFC3339) for point-in-time recovery; empty means latest. ## @field {string} oldName - Previous cluster name before deletion. +## @field {string} [serverName] - Barman server name (S3 path prefix) used by the original cluster when writing backups. Set this only when the original cluster had an explicit barmanObjectStore.serverName that differed from its Kubernetes resource name. ## @param {Bootstrap} bootstrap - Bootstrap configuration. bootstrap: @@ -161,3 +162,4 @@ bootstrap: # example: 2020-11-26 15:22:00.00000+00 recoveryTime: "" oldName: "" + serverName: "" diff --git a/packages/apps/tenant/Makefile b/packages/apps/tenant/Makefile index 3fe3810d..2f51d836 100644 --- a/packages/apps/tenant/Makefile +++ b/packages/apps/tenant/Makefile @@ -3,3 +3,6 @@ include ../../../hack/package.mk generate: cozyvalues-gen -m 'tenant' -v values.yaml -s values.schema.json -r README.md -g ../../../api/apps/v1alpha1/tenant/types.go ../../../hack/update-crd.sh + +test: + helm unittest . diff --git a/packages/apps/tenant/templates/cilium-lb-pool.yaml b/packages/apps/tenant/templates/cilium-lb-pool.yaml new file mode 100644 index 00000000..63dbcaca --- /dev/null +++ b/packages/apps/tenant/templates/cilium-lb-pool.yaml @@ -0,0 +1,28 @@ +{{- $exposeMode := (index .Values._cluster "expose-mode") | default "externalIPs" }} +{{- $exposeIngress := (index .Values._cluster "expose-ingress") | default "tenant-root" }} +{{- $exposeIPs := (index .Values._cluster "expose-external-ips") | default "" | nospace }} +{{- $ipsList := list }} +{{- range splitList "," $exposeIPs }} + {{- $ip := . | trim }} + {{- if $ip }} + {{- $ipsList = append $ipsList $ip }} + {{- end }} +{{- end }} +{{- $isPublishingIngressLB := and + (eq $exposeMode "loadBalancer") + (eq $exposeIngress .Release.Namespace) + .Values.ingress }} +{{- if and $isPublishingIngressLB $ipsList }} +apiVersion: cilium.io/v2 +kind: CiliumLoadBalancerIPPool +metadata: + name: {{ trimPrefix "tenant-" .Release.Namespace }}-exposure +spec: + blocks: + {{- range $ipsList }} + - cidr: {{ . }}{{ if not (contains "/" .) }}/{{ if contains ":" . }}128{{ else }}32{{ end }}{{ end }} + {{- end }} + serviceSelector: + matchLabels: + "io.kubernetes.service.namespace": {{ .Release.Namespace | quote }} +{{- end }} diff --git a/packages/apps/tenant/tests/exposure_test.yaml b/packages/apps/tenant/tests/exposure_test.yaml new file mode 100644 index 00000000..e86fab5b --- /dev/null +++ b/packages/apps/tenant/tests/exposure_test.yaml @@ -0,0 +1,141 @@ +suite: tenant CiliumLoadBalancerIPPool rendering for publishing.exposure=loadBalancer +templates: + - templates/cilium-lb-pool.yaml + +release: + name: tenant-root + namespace: tenant-root + +tests: + - it: default exposure (externalIPs) renders no pool + set: + ingress: true + _cluster: + expose-ingress: tenant-root + expose-external-ips: "192.0.2.10,192.0.2.11" + asserts: + - hasDocuments: + count: 0 + + - it: loadBalancer mode in publishing tenant with ingress=true renders v2 pool with namespace-only selector + set: + ingress: true + _cluster: + expose-ingress: tenant-root + expose-external-ips: "192.0.2.10,192.0.2.11" + expose-mode: loadBalancer + asserts: + - hasDocuments: + count: 1 + - equal: + path: apiVersion + value: cilium.io/v2 + - equal: + path: kind + value: CiliumLoadBalancerIPPool + - equal: + path: metadata.name + value: root-exposure + - equal: + path: spec.blocks + value: + - cidr: 192.0.2.10/32 + - cidr: 192.0.2.11/32 + - equal: + path: spec.serviceSelector.matchLabels["io.kubernetes.service.namespace"] + value: tenant-root + - notExists: + path: spec.serviceSelector.matchLabels["app.kubernetes.io/name"] + + - it: loadBalancer mode with IPv6 emits /128 CIDR + set: + ingress: true + _cluster: + expose-ingress: tenant-root + expose-external-ips: "2001:db8::1" + expose-mode: loadBalancer + asserts: + - equal: + path: spec.blocks + value: + - cidr: 2001:db8::1/128 + + - it: loadBalancer mode with mixed IPv4 and IPv6 emits correct CIDR per family + set: + ingress: true + _cluster: + expose-ingress: tenant-root + expose-external-ips: "192.0.2.10,2001:db8::1" + expose-mode: loadBalancer + asserts: + - equal: + path: spec.blocks + value: + - cidr: 192.0.2.10/32 + - cidr: 2001:db8::1/128 + + - it: loadBalancer mode accepts pre-CIDR input without double-suffixing + set: + ingress: true + _cluster: + expose-ingress: tenant-root + expose-external-ips: "192.0.2.10/32,2001:db8::1/128" + expose-mode: loadBalancer + asserts: + - equal: + path: spec.blocks + value: + - cidr: 192.0.2.10/32 + - cidr: 2001:db8::1/128 + + - it: loadBalancer mode filters out empty entries from externalIPs (trailing, leading, repeated commas) + set: + ingress: true + _cluster: + expose-ingress: tenant-root + expose-external-ips: "192.0.2.10,,192.0.2.11," + expose-mode: loadBalancer + asserts: + - hasDocuments: + count: 1 + - equal: + path: spec.blocks + value: + - cidr: 192.0.2.10/32 + - cidr: 192.0.2.11/32 + + - it: loadBalancer mode with ingress=false in publishing tenant renders no pool + set: + ingress: false + _cluster: + expose-ingress: tenant-root + expose-external-ips: "192.0.2.10" + expose-mode: loadBalancer + asserts: + - hasDocuments: + count: 0 + + - it: loadBalancer mode in a non-publishing tenant renders no pool + set: + ingress: true + _cluster: + expose-ingress: tenant-root + expose-external-ips: "192.0.2.10" + expose-mode: loadBalancer + release: + name: tenant-u1 + namespace: tenant-u1 + asserts: + - hasDocuments: + count: 0 + + - it: loadBalancer mode in publishing tenant with empty externalIPs renders no pool + set: + ingress: true + _cluster: + expose-ingress: tenant-root + expose-external-ips: "" + expose-mode: loadBalancer + asserts: + - hasDocuments: + count: 0 diff --git a/packages/apps/vm-instance/README.md b/packages/apps/vm-instance/README.md index a2b6603e..9d6a52b2 100644 --- a/packages/apps/vm-instance/README.md +++ b/packages/apps/vm-instance/README.md @@ -36,31 +36,32 @@ virtctl ssh @ ### Common parameters -| Name | Description | Type | Value | -| ------------------- | --------------------------------------------------------------------------------------------------------------------------------- | ---------- | ----------- | -| `external` | Enable external access from outside the cluster. | `bool` | `false` | -| `externalMethod` | Method to pass through traffic to the VM. | `string` | `PortList` | -| `externalPorts` | Ports to forward from outside the cluster. | `[]int` | `[22]` | -| `runStrategy` | Requested running state of the VirtualMachineInstance | `string` | `Always` | -| `instanceType` | Virtual Machine instance type. | `string` | `u1.medium` | -| `instanceProfile` | Virtual Machine preferences profile. | `string` | `ubuntu` | -| `disks` | List of disks to attach. | `[]object` | `[]` | -| `disks[i].name` | Disk name. | `string` | `""` | -| `disks[i].bus` | Disk bus type (e.g. "sata"). | `string` | `""` | -| `networks` | Networks to attach the VM to. | `[]object` | `[]` | -| `networks[i].name` | Network attachment name. | `string` | `""` | -| `subnets` | Deprecated: use networks instead. | `[]object` | `[]` | -| `subnets[i].name` | Network attachment name. | `string` | `""` | -| `gpus` | List of GPUs to attach (NVIDIA driver requires at least 4 GiB RAM). | `[]object` | `[]` | -| `gpus[i].name` | The name of the GPU resource to attach. | `string` | `""` | -| `cpuModel` | Model specifies the CPU model inside the VMI. List of available models https://github.com/libvirt/libvirt/tree/master/src/cpu_map | `string` | `""` | -| `resources` | Resource configuration for the virtual machine. | `object` | `{}` | -| `resources.cpu` | Number of CPU cores allocated. | `quantity` | `""` | -| `resources.memory` | Amount of memory allocated. | `quantity` | `""` | -| `resources.sockets` | Number of CPU sockets (vCPU topology). | `quantity` | `""` | -| `sshKeys` | List of SSH public keys for authentication. | `[]string` | `[]` | -| `cloudInit` | Cloud-init user data. | `string` | `""` | -| `cloudInitSeed` | Seed string to generate SMBIOS UUID for the VM. | `string` | `""` | +| Name | Description | Type | Value | +| ------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------- | ----------- | +| `external` | Enable external access from outside the cluster. | `bool` | `false` | +| `externalMethod` | Method to pass through traffic to the VM. | `string` | `PortList` | +| `externalPorts` | Ports to forward from outside the cluster. | `[]int` | `[22]` | +| `externalAllowICMP` | Whether to accept ICMP traffic to the VM in PortList mode (preserves ping and PMTU discovery). No effect in WholeIP mode. Default true so ping behaves as users expect even when port filtering is in effect. | `bool` | `true` | +| `runStrategy` | Requested running state of the VirtualMachineInstance | `string` | `Always` | +| `instanceType` | Virtual Machine instance type. | `string` | `u1.medium` | +| `instanceProfile` | Virtual Machine preferences profile. | `string` | `ubuntu` | +| `disks` | List of disks to attach. | `[]object` | `[]` | +| `disks[i].name` | Disk name. | `string` | `""` | +| `disks[i].bus` | Disk bus type (e.g. "sata"). | `string` | `""` | +| `networks` | Networks to attach the VM to. | `[]object` | `[]` | +| `networks[i].name` | Network attachment name. | `string` | `""` | +| `subnets` | Deprecated: use networks instead. | `[]object` | `[]` | +| `subnets[i].name` | Network attachment name. | `string` | `""` | +| `gpus` | List of GPUs to attach (NVIDIA driver requires at least 4 GiB RAM). | `[]object` | `[]` | +| `gpus[i].name` | The name of the GPU resource to attach. | `string` | `""` | +| `cpuModel` | Model specifies the CPU model inside the VMI. List of available models https://github.com/libvirt/libvirt/tree/master/src/cpu_map | `string` | `""` | +| `resources` | Resource configuration for the virtual machine. | `object` | `{}` | +| `resources.cpu` | Number of CPU cores allocated. | `quantity` | `""` | +| `resources.memory` | Amount of memory allocated. | `quantity` | `""` | +| `resources.sockets` | Number of CPU sockets (vCPU topology). | `quantity` | `""` | +| `sshKeys` | List of SSH public keys for authentication. | `[]string` | `[]` | +| `cloudInit` | Cloud-init user data. | `string` | `""` | +| `cloudInitSeed` | Seed string to generate SMBIOS UUID for the VM. | `string` | `""` | ## U Series diff --git a/packages/apps/vm-instance/templates/service.yaml b/packages/apps/vm-instance/templates/service.yaml index b7e3a420..cfeffa81 100644 --- a/packages/apps/vm-instance/templates/service.yaml +++ b/packages/apps/vm-instance/templates/service.yaml @@ -9,7 +9,10 @@ metadata: {{- if .Values.external }} service.kubernetes.io/service-proxy-name: "cozy-proxy" annotations: - networking.cozystack.io/wholeIP: "true" + networking.cozystack.io/wholeIP: {{ ternary "true" "false" (eq .Values.externalMethod "WholeIP") | quote }} + {{- if eq .Values.externalMethod "PortList" }} + networking.cozystack.io/allowICMP: {{ ternary "true" "false" (ne .Values.externalAllowICMP false) | quote }} + {{- end }} {{- end }} spec: type: {{ ternary "LoadBalancer" "ClusterIP" .Values.external }} diff --git a/packages/apps/vm-instance/values.schema.json b/packages/apps/vm-instance/values.schema.json index 34f7f634..01bd30a9 100644 --- a/packages/apps/vm-instance/values.schema.json +++ b/packages/apps/vm-instance/values.schema.json @@ -26,6 +26,11 @@ "type": "integer" } }, + "externalAllowICMP": { + "description": "Whether to accept ICMP traffic to the VM in PortList mode (preserves ping and PMTU discovery). No effect in WholeIP mode. Default true so ping behaves as users expect even when port filtering is in effect.", + "type": "boolean", + "default": true + }, "runStrategy": { "description": "Requested running state of the VirtualMachineInstance", "type": "string", diff --git a/packages/apps/vm-instance/values.yaml b/packages/apps/vm-instance/values.yaml index f07b75d3..92e399c2 100644 --- a/packages/apps/vm-instance/values.yaml +++ b/packages/apps/vm-instance/values.yaml @@ -31,6 +31,9 @@ externalMethod: PortList externalPorts: - 22 +## @param {bool} externalAllowICMP - Whether to accept ICMP traffic to the VM in PortList mode (preserves ping and PMTU discovery). No effect in WholeIP mode. Default true so ping behaves as users expect even when port filtering is in effect. +externalAllowICMP: true + ## @enum {string} RunStrategy - Requested running state of the VirtualMachineInstance ## @value Always - VMI should always be running ## @value Halted - VMI should never be running diff --git a/packages/core/platform/sources/hami.yaml b/packages/core/platform/sources/hami.yaml new file mode 100644 index 00000000..0184e405 --- /dev/null +++ b/packages/core/platform/sources/hami.yaml @@ -0,0 +1,24 @@ +--- +apiVersion: cozystack.io/v1alpha1 +kind: PackageSource +metadata: + name: cozystack.hami +spec: + sourceRef: + kind: OCIRepository + name: cozystack-packages + namespace: cozy-system + path: / + variants: + - name: default + dependsOn: + - cozystack.gpu-operator + components: + - name: hami + path: system/hami + valuesFiles: + - values.yaml + install: + privileged: true + namespace: cozy-hami + releaseName: hami diff --git a/packages/core/platform/sources/kubernetes-application.yaml b/packages/core/platform/sources/kubernetes-application.yaml index 9787cb19..088383ad 100644 --- a/packages/core/platform/sources/kubernetes-application.yaml +++ b/packages/core/platform/sources/kubernetes-application.yaml @@ -52,6 +52,8 @@ spec: path: system/cilium - name: kubernetes-gpu-operator path: system/gpu-operator + - name: kubernetes-hami + path: system/hami - name: kubernetes-vertical-pod-autoscaler path: system/vertical-pod-autoscaler - name: kubernetes-prometheus-operator-crds diff --git a/packages/core/platform/templates/apps.yaml b/packages/core/platform/templates/apps.yaml index 7ee0de52..6e1ae25a 100644 --- a/packages/core/platform/templates/apps.yaml +++ b/packages/core/platform/templates/apps.yaml @@ -30,6 +30,7 @@ stringData: expose-services: {{ .Values.publishing.exposedServices | join "," | quote }} expose-ingress: {{ .Values.publishing.ingressName | quote }} expose-external-ips: {{ .Values.publishing.externalIPs | join "," | quote }} + expose-mode: {{ .Values.publishing.exposure | default "externalIPs" | quote }} cluster-domain: {{ .Values.networking.clusterDomain | quote }} api-server-endpoint: {{ .Values.publishing.apiServerEndpoint | quote }} {{- with .Values.branding }} diff --git a/packages/core/platform/templates/bundles/iaas.yaml b/packages/core/platform/templates/bundles/iaas.yaml index ced0a322..66ec98b2 100644 --- a/packages/core/platform/templates/bundles/iaas.yaml +++ b/packages/core/platform/templates/bundles/iaas.yaml @@ -10,6 +10,7 @@ {{include "cozystack.platform.package.default" (list "cozystack.kubevirt-cdi" $) }} {{include "cozystack.platform.package.optional.default" (list "cozystack.vm-default-images" $) }} {{include "cozystack.platform.package.optional.default" (list "cozystack.gpu-operator" $) }} +{{include "cozystack.platform.package.optional.default" (list "cozystack.hami" $) }} {{include "cozystack.platform.package.default" (list "cozystack.kamaji" $) }} {{include "cozystack.platform.package.default" (list "cozystack.capi-operator" $) }} {{include "cozystack.platform.package.default" (list "cozystack.capi-provider-bootstrap-kubeadm" $) }} diff --git a/packages/core/platform/templates/bundles/system.yaml b/packages/core/platform/templates/bundles/system.yaml index 43ea7266..6c587c82 100644 --- a/packages/core/platform/templates/bundles/system.yaml +++ b/packages/core/platform/templates/bundles/system.yaml @@ -105,9 +105,6 @@ {{- /* cozystack-api DaemonSet */ -}} {{- $apiValues := dict "cozystackAPI" (dict "nodeSelector" $genericNodeSelector) -}} {{- $_ := set $cozystackEngineComponents "cozystack-api" (dict "values" $apiValues) -}} -{{- /* lineage-controller-webhook DaemonSet */ -}} -{{- $lineageValues := dict "lineageControllerWebhook" (dict "nodeSelector" $genericNodeSelector) -}} -{{- $_ := set $cozystackEngineComponents "lineage-controller-webhook" (dict "values" $lineageValues) -}} {{- end -}} {{- if .Values.authentication.oidc.enabled }} {{include "cozystack.platform.package" (list "cozystack.cozystack-engine" "oidc" $ $cozystackEngineComponents) }} diff --git a/packages/core/platform/values.yaml b/packages/core/platform/values.yaml index f33926db..605e23bc 100644 --- a/packages/core/platform/values.yaml +++ b/packages/core/platform/values.yaml @@ -45,6 +45,41 @@ publishing: - cdi-uploadproxy apiServerEndpoint: "" # example: "https://api.example.org" externalIPs: [] + # Exposure mode for the ingress-nginx Service. When "externalIPs" (current + # default) is selected, the Service is created as ClusterIP with + # Service.spec.externalIPs set from publishing.externalIPs. When + # "loadBalancer" is selected, the Service is type: LoadBalancer and a + # CiliumLoadBalancerIPPool makes those same addresses allocatable via LB IPAM. + # + # Service.spec.externalIPs is deprecated upstream in Kubernetes v1.36 + # (KEP-5707). The AllowServiceExternalIPs feature gate is expected to default + # to false around v1.40 and the implementation removed around v1.43 — switch + # to "loadBalancer" before upgrading past v1.40. + # + # Caveats for the "loadBalancer" mode: + # - publishing.externalIPs must contain at least one non-empty address, + # otherwise the chart render fails with an explicit error (a LoadBalancer + # Service without a pool would sit in forever). + # - The ingress-nginx Service is created with externalTrafficPolicy: Local + # to preserve the client source IP. Traffic arriving on a node that does + # not host an ingress-nginx pod is dropped, so the external IP must be + # routed to a node that runs the ingress pod (floating IP / keepalived / + # upstream router / podAntiAffinity). + # - Cilium does NOT announce the IP on its own unless L2 announcements or + # BGP are enabled in the Cilium values (disabled by default in Cozystack). + # This mode assumes the operator already routes the externalIPs to a + # cluster node; enabling announcements is out of scope for this setting. + # - Switching this value on a running cluster causes the ingress-nginx + # Service to be recreated (the HelmRelease has upgrade.force: true and + # the Service kind changes between ClusterIP and LoadBalancer). Expect a + # brief interruption of ingress traffic during the flip. + # + # Scope: this setting only controls the ingress-nginx Service. Other + # cozystack components that currently write Service.spec.externalIPs directly + # (e.g. the vpn app at packages/apps/vpn/templates/service.yaml) are NOT + # migrated by flipping this value and must be addressed separately before + # the AllowServiceExternalIPs feature gate flips to off in ~v1.40. + exposure: externalIPs # "externalIPs" or "loadBalancer" certificates: solver: http01 # "http01" or "dns01" issuerName: letsencrypt-prod diff --git a/packages/extra/etcd/Makefile b/packages/extra/etcd/Makefile index f37d6e1e..2b3ed61e 100644 --- a/packages/extra/etcd/Makefile +++ b/packages/extra/etcd/Makefile @@ -5,3 +5,6 @@ include ../../../hack/package.mk generate: cozyvalues-gen -v values.yaml -s values.schema.json -r README.md ../../../hack/update-crd.sh + +test: + helm unittest . diff --git a/packages/extra/etcd/templates/hook/job.yaml b/packages/extra/etcd/templates/hook/job.yaml deleted file mode 100644 index 3bf2f84b..00000000 --- a/packages/extra/etcd/templates/hook/job.yaml +++ /dev/null @@ -1,39 +0,0 @@ -{{- $shouldUpdateCerts := true }} -{{- $configMap := lookup "v1" "ConfigMap" .Release.Namespace "etcd-deployed-version" }} -{{- if $configMap }} - {{- $deployedVersion := index $configMap "data" "version" }} - {{- if $deployedVersion | semverCompare ">= 2.6.1" }} - {{- $shouldUpdateCerts = false }} - {{- end }} -{{- end }} - -{{- if $shouldUpdateCerts }} ---- -apiVersion: batch/v1 -kind: Job -metadata: - name: etcd-hook - annotations: - helm.sh/hook: post-upgrade - helm.sh/hook-delete-policy: before-hook-creation,hook-succeeded -spec: - template: - metadata: - labels: - policy.cozystack.io/allow-to-apiserver: "true" - spec: - serviceAccountName: etcd-hook - containers: - - name: kubectl - image: docker.io/alpine/k8s:1.33.4 - command: - - sh - args: - - -exc - - |- - kubectl --namespace={{ .Release.Namespace }} delete secrets etcd-ca-tls etcd-peer-ca-tls - sleep 10 - kubectl --namespace={{ .Release.Namespace }} delete secrets etcd-client-tls etcd-peer-tls etcd-server-tls - kubectl --namespace={{ .Release.Namespace }} delete pods --selector=app.kubernetes.io/instance=etcd,app.kubernetes.io/managed-by=etcd-operator,app.kubernetes.io/name=etcd,cozystack.io/service=etcd - restartPolicy: Never -{{- end }} diff --git a/packages/extra/etcd/templates/hook/role.yaml b/packages/extra/etcd/templates/hook/role.yaml deleted file mode 100644 index 327eeadb..00000000 --- a/packages/extra/etcd/templates/hook/role.yaml +++ /dev/null @@ -1,26 +0,0 @@ -apiVersion: rbac.authorization.k8s.io/v1 -kind: Role -metadata: - annotations: - helm.sh/hook: post-upgrade - helm.sh/hook-delete-policy: before-hook-creation,hook-succeeded - name: etcd-hook -rules: -- apiGroups: - - "" - resources: - - secrets - - pods - verbs: - - get - - list - - watch - - delete -- apiGroups: - - "" - resources: - - configmaps - verbs: - - get - - list - - watch diff --git a/packages/extra/etcd/templates/hook/rolebinding.yaml b/packages/extra/etcd/templates/hook/rolebinding.yaml deleted file mode 100644 index 0ee0ffd1..00000000 --- a/packages/extra/etcd/templates/hook/rolebinding.yaml +++ /dev/null @@ -1,15 +0,0 @@ -apiVersion: rbac.authorization.k8s.io/v1 -kind: RoleBinding -metadata: - name: etcd-hook - annotations: - helm.sh/hook: post-upgrade - helm.sh/hook-delete-policy: before-hook-creation,hook-succeeded -roleRef: - apiGroup: rbac.authorization.k8s.io - kind: Role - name: etcd-hook -subjects: - - kind: ServiceAccount - name: etcd-hook - namespace: {{ .Release.Namespace | quote }} diff --git a/packages/extra/etcd/templates/hook/serviceaccount.yaml b/packages/extra/etcd/templates/hook/serviceaccount.yaml deleted file mode 100644 index 552fb5fc..00000000 --- a/packages/extra/etcd/templates/hook/serviceaccount.yaml +++ /dev/null @@ -1,7 +0,0 @@ -apiVersion: v1 -kind: ServiceAccount -metadata: - name: etcd-hook - annotations: - helm.sh/hook: post-upgrade - helm.sh/hook-delete-policy: before-hook-creation,hook-succeeded diff --git a/packages/extra/etcd/templates/version.yaml b/packages/extra/etcd/templates/version.yaml deleted file mode 100644 index cc9375bb..00000000 --- a/packages/extra/etcd/templates/version.yaml +++ /dev/null @@ -1,6 +0,0 @@ -apiVersion: v1 -kind: ConfigMap -metadata: - name: etcd-deployed-version -data: - version: {{ .Chart.Version }} diff --git a/packages/extra/etcd/tests/no-post-upgrade-hook_test.yaml b/packages/extra/etcd/tests/no-post-upgrade-hook_test.yaml new file mode 100644 index 00000000..0e4f5aa3 --- /dev/null +++ b/packages/extra/etcd/tests/no-post-upgrade-hook_test.yaml @@ -0,0 +1,34 @@ +suite: etcd chart does not ship a destructive post-upgrade cert-regeneration hook + +release: + name: etcd + namespace: tenant-root + +templates: + - templates/check-release-name.yaml + - templates/dashboard-resourcemap.yaml + - templates/datastore.yaml + - templates/etcd-defrag.yaml + - templates/hook/job.yaml + - templates/podscrape.yaml + - templates/prometheus-rules.yaml + - templates/version.yaml + +tests: + - it: renders no Job named etcd-hook + documentSelector: + path: metadata.name + value: etcd-hook + skipEmptyTemplates: true + asserts: + - hasDocuments: + count: 0 + + - it: renders no ConfigMap named etcd-deployed-version + documentSelector: + path: metadata.name + value: etcd-deployed-version + skipEmptyTemplates: true + asserts: + - hasDocuments: + count: 0 diff --git a/packages/extra/ingress/Makefile b/packages/extra/ingress/Makefile index 958ce484..65b53c03 100644 --- a/packages/extra/ingress/Makefile +++ b/packages/extra/ingress/Makefile @@ -10,3 +10,6 @@ get-cloudflare-ips: generate: cozyvalues-gen -v values.yaml -s values.schema.json -r README.md ../../../hack/update-crd.sh + +test: + helm unittest . diff --git a/packages/extra/ingress/README.md b/packages/extra/ingress/README.md index 0e786dfb..c541d8e9 100644 --- a/packages/extra/ingress/README.md +++ b/packages/extra/ingress/README.md @@ -14,3 +14,17 @@ | `resources.memory` | Memory (RAM) available to each replica. | `quantity` | `""` | | `resourcesPreset` | Default sizing preset used when `resources` is omitted. | `string` | `micro` | + +## Exposure mode + +The ingress Service type is driven by the cluster-wide `publishing.exposure` value in the platform chart, not by any key in this package. Two modes exist: + +- `externalIPs` (default) has three rendered shapes: + - Release namespace matches `publishing.ingressName` AND `publishing.externalIPs` is non-empty → Service is `ClusterIP` with `Service.spec.externalIPs` set from that list and `externalTrafficPolicy: Cluster`. + - Release namespace matches `publishing.ingressName` but `publishing.externalIPs` is empty → Service falls back to `type: LoadBalancer` with `externalTrafficPolicy: Local`. + - Release namespace does not match `publishing.ingressName` (non-root tenants) → Service is `type: LoadBalancer` with `externalTrafficPolicy: Local`. + `Service.spec.externalIPs` is deprecated upstream in Kubernetes v1.36 (KEP-5707); plan migration before v1.40. +- `loadBalancer` — Service is `type: LoadBalancer` with `externalTrafficPolicy: Local`, and a `CiliumLoadBalancerIPPool` makes the addresses in `publishing.externalIPs` allocatable via Cilium LB IPAM. Requires `publishing.externalIPs` to contain at least one non-empty address (render fails otherwise) and assumes the addresses are already routed to a cluster node (floating IP / upstream router). See the inline comment on `publishing.exposure` in the platform chart for full caveats, including the note that switching the value on a running cluster causes the ingress Service to be recreated. + +This setting only migrates ingress-nginx away from `Service.spec.externalIPs`. Other cozystack components that use the same deprecated field (e.g. the `vpn` app) must be migrated separately before Kubernetes v1.40 flips the `AllowServiceExternalIPs` feature gate off. + diff --git a/packages/extra/ingress/templates/nginx-ingress.yaml b/packages/extra/ingress/templates/nginx-ingress.yaml index ca50d276..8e1f00fb 100644 --- a/packages/extra/ingress/templates/nginx-ingress.yaml +++ b/packages/extra/ingress/templates/nginx-ingress.yaml @@ -1,5 +1,19 @@ {{- $exposeIngress := (index .Values._cluster "expose-ingress") | default "tenant-root" }} {{- $exposeExternalIPs := (index .Values._cluster "expose-external-ips") | default "" | nospace }} +{{- $exposeMode := (index .Values._cluster "expose-mode") | default "externalIPs" }} +{{- $exposeIPsList := list }} +{{- range splitList "," $exposeExternalIPs }} + {{- $ip := . | trim }} + {{- if $ip }} + {{- $exposeIPsList = append $exposeIPsList $ip }} + {{- end }} +{{- end }} +{{- if not (has $exposeMode (list "externalIPs" "loadBalancer")) }} +{{- fail (printf "unknown publishing.exposure mode %q: must be \"externalIPs\" or \"loadBalancer\"" $exposeMode) }} +{{- end }} +{{- if and (eq $exposeMode "loadBalancer") (eq $exposeIngress .Release.Namespace) (not $exposeIPsList) }} +{{- fail "publishing.exposure=loadBalancer requires publishing.externalIPs to contain at least one non-empty address: the CiliumLoadBalancerIPPool has nothing to advertise and the Service would stay in ." }} +{{- end }} apiVersion: helm.toolkit.fluxcd.io/v2 kind: HelmRelease metadata: @@ -41,9 +55,12 @@ spec: enabled: false {{- end }} service: - {{- if and (eq $exposeIngress .Release.Namespace) $exposeExternalIPs }} + {{- if and (eq $exposeIngress .Release.Namespace) (eq $exposeMode "loadBalancer") }} + type: LoadBalancer + externalTrafficPolicy: Local + {{- else if and (eq $exposeIngress .Release.Namespace) $exposeIPsList }} externalIPs: - {{- toYaml (splitList "," $exposeExternalIPs) | nindent 12 }} + {{- toYaml $exposeIPsList | nindent 12 }} type: ClusterIP externalTrafficPolicy: Cluster {{- else }} diff --git a/packages/extra/ingress/tests/exposure_test.yaml b/packages/extra/ingress/tests/exposure_test.yaml new file mode 100644 index 00000000..1c76b3da --- /dev/null +++ b/packages/extra/ingress/tests/exposure_test.yaml @@ -0,0 +1,165 @@ +suite: ingress exposure modes +templates: + - templates/nginx-ingress.yaml + +release: + name: ingress + namespace: tenant-root + +tests: + - it: default exposure (externalIPs) renders ClusterIP Service with spec.externalIPs + set: + _cluster: + expose-ingress: tenant-root + expose-external-ips: "192.0.2.10,192.0.2.11" + asserts: + - template: templates/nginx-ingress.yaml + equal: + path: spec.values.ingress-nginx.controller.service.type + value: ClusterIP + - template: templates/nginx-ingress.yaml + equal: + path: spec.values.ingress-nginx.controller.service.externalTrafficPolicy + value: Cluster + - template: templates/nginx-ingress.yaml + equal: + path: spec.values.ingress-nginx.controller.service.externalIPs + value: + - 192.0.2.10 + - 192.0.2.11 + - template: templates/nginx-ingress.yaml + notExists: + path: spec.values.ingress-nginx.controller.service.labels + + - it: legacy config without expose-mode falls back to externalIPs behavior + set: + _cluster: + expose-ingress: tenant-root + expose-external-ips: "192.0.2.10" + asserts: + - template: templates/nginx-ingress.yaml + equal: + path: spec.values.ingress-nginx.controller.service.type + value: ClusterIP + + - it: externalIPs mode in a namespace other than publishing.ingressName renders LoadBalancer fallback without externalIPs + set: + _cluster: + expose-ingress: tenant-root + expose-external-ips: "192.0.2.10" + release: + namespace: tenant-other + asserts: + - template: templates/nginx-ingress.yaml + equal: + path: spec.values.ingress-nginx.controller.service.type + value: LoadBalancer + - template: templates/nginx-ingress.yaml + equal: + path: spec.values.ingress-nginx.controller.service.externalTrafficPolicy + value: Local + - template: templates/nginx-ingress.yaml + notExists: + path: spec.values.ingress-nginx.controller.service.externalIPs + + - it: loadBalancer mode renders LoadBalancer Service without externalIPs on the Service + set: + _cluster: + expose-ingress: tenant-root + expose-external-ips: "192.0.2.10,192.0.2.11" + expose-mode: loadBalancer + asserts: + - template: templates/nginx-ingress.yaml + equal: + path: spec.values.ingress-nginx.controller.service.type + value: LoadBalancer + - template: templates/nginx-ingress.yaml + equal: + path: spec.values.ingress-nginx.controller.service.externalTrafficPolicy + value: Local + - template: templates/nginx-ingress.yaml + notExists: + path: spec.values.ingress-nginx.controller.service.labels + - template: templates/nginx-ingress.yaml + notExists: + path: spec.values.ingress-nginx.controller.service.externalIPs + + - it: loadBalancer mode without externalIPs fails chart render with explicit message + set: + _cluster: + expose-ingress: tenant-root + expose-external-ips: "" + expose-mode: loadBalancer + asserts: + - template: templates/nginx-ingress.yaml + failedTemplate: + errorMessage: "publishing.exposure=loadBalancer requires publishing.externalIPs to contain at least one non-empty address: the CiliumLoadBalancerIPPool has nothing to advertise and the Service would stay in ." + + - it: unknown exposure mode is rejected with a clear error (case-sensitive enum) + set: + _cluster: + expose-ingress: tenant-root + expose-external-ips: "192.0.2.10" + expose-mode: LoadBalancer + asserts: + - template: templates/nginx-ingress.yaml + failedTemplate: + errorMessage: "unknown publishing.exposure mode \"LoadBalancer\": must be \"externalIPs\" or \"loadBalancer\"" + + - it: another typo in exposure mode also fails + set: + _cluster: + expose-ingress: tenant-root + expose-external-ips: "192.0.2.10" + expose-mode: loadbalancer + asserts: + - template: templates/nginx-ingress.yaml + failedTemplate: + errorMessage: "unknown publishing.exposure mode \"loadbalancer\": must be \"externalIPs\" or \"loadBalancer\"" + + - it: loadBalancer mode in a namespace other than publishing.ingressName falls back to LoadBalancer Service without pool + set: + _cluster: + expose-ingress: tenant-root + expose-external-ips: "192.0.2.10" + expose-mode: loadBalancer + release: + namespace: tenant-other + asserts: + - template: templates/nginx-ingress.yaml + equal: + path: spec.values.ingress-nginx.controller.service.type + value: LoadBalancer + - template: templates/nginx-ingress.yaml + equal: + path: spec.values.ingress-nginx.controller.service.externalTrafficPolicy + value: Local + - template: templates/nginx-ingress.yaml + notExists: + path: spec.values.ingress-nginx.controller.service.labels + - template: templates/nginx-ingress.yaml + notExists: + path: spec.values.ingress-nginx.controller.service.externalIPs + + - it: loadBalancer mode with only-empty externalIPs fails chart render (comma-only input) + set: + _cluster: + expose-ingress: tenant-root + expose-external-ips: ",," + expose-mode: loadBalancer + asserts: + - template: templates/nginx-ingress.yaml + failedTemplate: + errorMessage: "publishing.exposure=loadBalancer requires publishing.externalIPs to contain at least one non-empty address: the CiliumLoadBalancerIPPool has nothing to advertise and the Service would stay in ." + + - it: externalIPs mode also filters out empty entries (trailing comma) + set: + _cluster: + expose-ingress: tenant-root + expose-external-ips: "192.0.2.10," + asserts: + - template: templates/nginx-ingress.yaml + equal: + path: spec.values.ingress-nginx.controller.service.externalIPs + value: + - 192.0.2.10 diff --git a/packages/system/cozy-proxy/charts/cozy-proxy/Chart.yaml b/packages/system/cozy-proxy/charts/cozy-proxy/Chart.yaml index 13352680..58f166f6 100644 --- a/packages/system/cozy-proxy/charts/cozy-proxy/Chart.yaml +++ b/packages/system/cozy-proxy/charts/cozy-proxy/Chart.yaml @@ -2,5 +2,5 @@ apiVersion: v2 name: cozy-proxy description: A simple kube-proxy addon for 1:1 NAT services in Kubernetes using an NFT backend type: application -version: 0.2.0 -appVersion: 0.2.0 +version: 0.3.0 +appVersion: 0.3.0 diff --git a/packages/system/cozy-proxy/charts/cozy-proxy/values.yaml b/packages/system/cozy-proxy/charts/cozy-proxy/values.yaml index 8cde5bed..e143e926 100644 --- a/packages/system/cozy-proxy/charts/cozy-proxy/values.yaml +++ b/packages/system/cozy-proxy/charts/cozy-proxy/values.yaml @@ -1,6 +1,6 @@ image: repository: ghcr.io/cozystack/cozystack/cozy-proxy - tag: v0.2.0 + tag: v0.3.0 pullPolicy: IfNotPresent daemonset: diff --git a/packages/system/gpu-operator/examples/README.md b/packages/system/gpu-operator/examples/README.md new file mode 100644 index 00000000..b6f170c8 --- /dev/null +++ b/packages/system/gpu-operator/examples/README.md @@ -0,0 +1,121 @@ +# GPU operator — native pod workload on Talos (reference) + +The files in this directory are **not** templates. They are reference +artifacts that document one working configuration for running GPU +workloads directly in pods on a Talos-based Cozystack cluster, together +with the DCGM metrics needed by the `gpu/gpu-performance` Grafana +dashboard. + +The out-of-the-box `values-talos.yaml` for this package targets the +sandbox (VFIO passthrough to KubeVirt VMs) scenario. The files here +illustrate an alternative — running CUDA workloads in regular pods with +the NVIDIA device plugin — and the workarounds it currently requires on +Talos. + +## Files + +- [`values-native-talos.yaml`](./values-native-talos.yaml) — Cozystack + `Package` values that disable sandbox workloads, enable the device + plugin, point `hostPaths.driverInstallDir` at the staging location + used by the compat DaemonSet, and wire DCGM to the custom metrics + ConfigMap. +- [`dcgm-custom-metrics.yaml`](./dcgm-custom-metrics.yaml) — `ConfigMap` + with a DCGM metrics CSV that adds profiling, ECC, throttling and + energy counters on top of the upstream defaults. The CSV is a + superset needed for full coverage of the `gpu/gpu-performance` + dashboard. Which parts are actually required depends on which + dashboards you ship — see the table below. +- [`nvidia-driver-compat.yaml`](./nvidia-driver-compat.yaml) — DaemonSet + that stages `libnvidia-ml.so.1` and `nvidia-smi` from the Talos glibc + tree into a path where the NVIDIA GPU Operator validator expects + them. See the "Why the compat DaemonSet exists" section below. + +## Why these are reference, not templates + +Shipping these as first-class templates would silently impose +assumptions that do not hold for every user: + +- Whether the NVIDIA Talos system extension is installed on the nodes. +- Whether GPUs are exposed directly to pods or passed through to VMs. +- The exact path the installed driver ends up at (depends on the + extension version and Talos release). + +The sandbox-oriented `values-talos.yaml` remains the default. Operators +who want native pod GPU workloads can start from this directory and +adapt as needed. + +## Why the compat DaemonSet exists + +The NVIDIA GPU Operator validator checks for `libnvidia-ml.so.1` and +`bin/nvidia-smi` in the path given by `hostPaths.driverInstallDir`. +Talos installs them under `/usr/local/glibc/usr/lib/` and +`/usr/local/bin/`, which the validator does not look at. Until upstream +addresses [NVIDIA/gpu-operator#1687][1], the DaemonSet copies those +files into a directory the validator does inspect and creates the +`.driver-ctr-ready` flag file so the validator proceeds. + +[1]: https://github.com/NVIDIA/gpu-operator/issues/1687 + +The compat DaemonSet runs privileged and bind-mounts host paths, so +the target namespace must allow privileged pods. On clusters that +enforce the Kubernetes Pod Security Standards at `baseline` or +`restricted`, label the namespace with +`pod-security.kubernetes.io/enforce: privileged` (and the matching +`audit`/`warn` labels if the admission webhook is configured to +surface violations) before applying the manifest. + +## Dashboards and what DCGM metrics they need + +Five GPU dashboards live under `gpu/*` in +`packages/system/monitoring/dashboards-infra.list`. All of them share +`packages/system/monitoring-agents/alerts/gpu-recording.rules.yaml` as +their source of aggregated series. The recording rules are safe to +ship on any cluster — they evaluate to empty series when DCGM is not +scraped, or when optional counters are missing. + +What each dashboard needs on top of the upstream DCGM Exporter +[`default-counters.csv`][default-csv]: + +| Dashboard | Scope | Needs beyond defaults | +| ----------------- | ---------------------------------- | ----------------------------------------------------------------------- | +| `gpu-performance` | Per-node, per-GPU deep dive | `DCGM_FI_DEV_POWER_VIOLATION`, `DCGM_FI_DEV_THERMAL_VIOLATION` | +| `gpu-efficiency` | Per-workload util vs tensor active | `DCGM_FI_DEV_POWER_VIOLATION`, `DCGM_FI_DEV_THERMAL_VIOLATION` (via `gpu:*_throttle_fraction:rate5m` recording rules) | +| `gpu-fleet` | Cluster-wide admin inventory | `DCGM_FI_DEV_POWER_MGMT_LIMIT` (for the TDP vs draw panel) | +| `gpu-quotas` | Kube-quota vs live usage | `kube_pod_container_resource_requests`, `kube_pod_status_phase`, `kube_node_status_allocatable` (via `namespace:gpu_count:allocated` / `cluster:gpu_count:*` recording rules) | +| `gpu-tenants` | Per-namespace tenant view | nothing (works on default counters) | + +`DCGM_FI_PROF_PIPE_TENSOR_ACTIVE` and `DCGM_FI_PROF_GR_ENGINE_ACTIVE` +are already in the upstream default set for the pinned DCGM Exporter +version, so the tensor-saturation and engine-active panels work without +any CSV override. The three counters listed in the table — throttling +violations and the power management limit — are the only extras the +tracked dashboards need. The recording rules in +`gpu-recording.rules.yaml` consume utilization, FB used, power, +temperature and the tensor-active profiling counter from the default +set, plus `DCGM_FI_DEV_POWER_VIOLATION` and +`DCGM_FI_DEV_THERMAL_VIOLATION` — used by the +`gpu.recording.efficiency.1m` group to derive the +`gpu:power_throttle_fraction:rate5m` and +`gpu:thermal_throttle_fraction:rate5m` series consumed by the +throttling panels on the efficiency and fleet dashboards. + +The `gpu.recording.throttle.validation.5m` group additionally ships the +`GPUThrottleFractionOverOne` alert (severity `warning`) as a regression +detector: it fires when either throttle-fraction series exceeds 1.0, +which would indicate that DCGM changed the scale/divisor of the +underlying violation counters and the recording rules need to be +re-derived. + +## Verification status + +The minimum-CSV claims above are verified by +`hack/check-gpu-recording-rules.bats`, which cross-checks every +`DCGM_FI_*` reference in the tracked GPU dashboards and recording rules +against the union of the upstream default set (snapshotted at +`hack/dcgm-default-counters.csv` for the pinned DCGM Exporter version) +and the custom CSV in `dcgm-custom-metrics.yaml`. When the DCGM +Exporter image in `packages/system/gpu-operator/charts/gpu-operator/values.yaml` +is bumped, refresh the snapshot from the matching tag of the +[`NVIDIA/dcgm-exporter`][default-csv] repository and rerun the test. + +[default-csv]: https://github.com/NVIDIA/dcgm-exporter/blob/main/etc/default-counters.csv diff --git a/packages/system/gpu-operator/examples/dcgm-custom-metrics.yaml b/packages/system/gpu-operator/examples/dcgm-custom-metrics.yaml new file mode 100644 index 00000000..ef2b0b88 --- /dev/null +++ b/packages/system/gpu-operator/examples/dcgm-custom-metrics.yaml @@ -0,0 +1,88 @@ +# Custom DCGM Exporter metrics CSV. Referenced from +# examples/values-native-talos.yaml via dcgmExporter.config.name. +# +# Extends the upstream default set with profiling counters, ECC, page +# retirement, row remap, energy and throttling violations — everything +# the gpu/gpu-performance dashboard and the GPU recording rules in +# monitoring-agents expect. +apiVersion: v1 +kind: ConfigMap +metadata: + name: dcgm-custom-metrics + namespace: cozy-gpu-operator +data: + dcgm-metrics.csv: | + # Format + # If line starts with a '#' it is considered a comment + # DCGM FIELD, Prometheus metric type, help message + + # Identity + DCGM_FI_DRIVER_VERSION, label, Driver version. + + # Clocks + DCGM_FI_DEV_SM_CLOCK, gauge, SM clock frequency (in MHz). + DCGM_FI_DEV_MEM_CLOCK, gauge, Memory clock frequency (in MHz). + + # Temperature + DCGM_FI_DEV_MEMORY_TEMP, gauge, Memory temperature (in C). + DCGM_FI_DEV_GPU_TEMP, gauge, GPU temperature (in C). + + # Power + DCGM_FI_DEV_POWER_USAGE, gauge, Power draw (in W). + DCGM_FI_DEV_POWER_MGMT_LIMIT, gauge, Current power management limit (in W). + DCGM_FI_DEV_TOTAL_ENERGY_CONSUMPTION, counter, Total energy consumption since boot (in mJ). + + # PCIE + DCGM_FI_DEV_PCIE_REPLAY_COUNTER, counter, Total number of PCIe retries. + + # Utilization (the sample period varies depending on the product) + DCGM_FI_DEV_GPU_UTIL, gauge, GPU utilization (in %). + DCGM_FI_DEV_MEM_COPY_UTIL, gauge, Memory utilization (in %). + DCGM_FI_DEV_ENC_UTIL, gauge, Encoder utilization (in %). + DCGM_FI_DEV_DEC_UTIL, gauge, Decoder utilization (in %). + + # Errors and violations + DCGM_FI_DEV_XID_ERRORS, gauge, Value of the last XID error encountered. + + # Memory usage + DCGM_FI_DEV_FB_FREE, gauge, Framebuffer memory free (in MiB). + DCGM_FI_DEV_FB_USED, gauge, Framebuffer memory used (in MiB). + DCGM_FI_DEV_FB_RESERVED, gauge, Framebuffer memory reserved (in MiB). + + # ECC (supported on datacenter-class GPUs such as A10) + DCGM_FI_DEV_ECC_SBE_VOL_TOTAL, counter, Total number of single-bit volatile ECC errors. + DCGM_FI_DEV_ECC_DBE_VOL_TOTAL, counter, Total number of double-bit volatile ECC errors. + DCGM_FI_DEV_ECC_SBE_AGG_TOTAL, counter, Total number of single-bit persistent ECC errors. + DCGM_FI_DEV_ECC_DBE_AGG_TOTAL, counter, Total number of double-bit persistent ECC errors. + + # Retired pages + DCGM_FI_DEV_RETIRED_SBE, counter, Total number of retired pages due to single-bit errors. + DCGM_FI_DEV_RETIRED_DBE, counter, Total number of retired pages due to double-bit errors. + DCGM_FI_DEV_RETIRED_PENDING, counter, Total number of pages pending retirement. + + # Row remapping (not applicable to GDDR6 but left for datacenter GPUs) + DCGM_FI_DEV_UNCORRECTABLE_REMAPPED_ROWS, counter, Number of remapped rows for uncorrectable errors. + DCGM_FI_DEV_CORRECTABLE_REMAPPED_ROWS, counter, Number of remapped rows for correctable errors. + DCGM_FI_DEV_ROW_REMAP_FAILURE, gauge, Whether remapping of rows has failed. + + # Throttle / violation counters (crucial for SLA and tenant monitoring) + DCGM_FI_DEV_POWER_VIOLATION, counter, Throttling duration due to power constraints (us per docs; ns on DCGM 3.x). + DCGM_FI_DEV_THERMAL_VIOLATION, counter, Throttling duration due to thermal constraints (us per docs; ns on DCGM 3.x). + DCGM_FI_DEV_SYNC_BOOST_VIOLATION, counter, Throttling duration due to sync-boost constraints (us per docs; ns on DCGM 3.x). + DCGM_FI_DEV_BOARD_LIMIT_VIOLATION, counter, Throttling duration due to board limit constraints (us per docs; ns on DCGM 3.x). + DCGM_FI_DEV_LOW_UTIL_VIOLATION, counter, Throttling duration due to low utilization (us per docs; ns on DCGM 3.x). + DCGM_FI_DEV_RELIABILITY_VIOLATION, counter, Throttling duration due to reliability constraints (us per docs; ns on DCGM 3.x). + + # NVLink — DCGM silently drops this metric on GPUs without NVLink, so + # enabling it unconditionally is safe and keeps this file reusable + # across GPU families. + DCGM_FI_DEV_NVLINK_BANDWIDTH_TOTAL, counter, Total number of NVLink bandwidth counters for all lanes. + + # DCP (profiling) metrics — supported from Ampere onwards. + DCGM_FI_PROF_GR_ENGINE_ACTIVE, gauge, Ratio of time the graphics engine is active. + DCGM_FI_PROF_SM_ACTIVE, gauge, The ratio of cycles an SM has at least 1 warp assigned. + DCGM_FI_PROF_SM_OCCUPANCY, gauge, The ratio of number of warps resident on an SM. + DCGM_FI_PROF_PIPE_TENSOR_ACTIVE, gauge, Ratio of cycles the tensor (HMMA) pipe is active. + DCGM_FI_PROF_DRAM_ACTIVE, gauge, Ratio of cycles the device memory interface is active sending or receiving data. + DCGM_FI_PROF_PCIE_TX_BYTES, counter, The number of bytes of active PCIe tx (transmit) data including both header and payload. + DCGM_FI_PROF_PCIE_RX_BYTES, counter, The number of bytes of active PCIe rx (read) data including both header and payload. diff --git a/packages/system/gpu-operator/examples/nvidia-driver-compat.yaml b/packages/system/gpu-operator/examples/nvidia-driver-compat.yaml new file mode 100644 index 00000000..fc9d5981 --- /dev/null +++ b/packages/system/gpu-operator/examples/nvidia-driver-compat.yaml @@ -0,0 +1,109 @@ +# Workaround for https://github.com/NVIDIA/gpu-operator/issues/1687 +# +# On Talos, the NVIDIA system extension installs driver files under +# /usr/local/glibc/usr/lib/ and /usr/local/bin/. The GPU Operator +# validator only looks for libnvidia-ml.so.1 and bin/nvidia-smi under +# the path configured as hostPaths.driverInstallDir. This DaemonSet +# stages the required files into /var/nvidia-driver on each node and +# creates the .driver-ctr-ready flag so the validator proceeds. +# +# Paired with examples/values-native-talos.yaml, which sets +# hostPaths.driverInstallDir to /var/nvidia-driver. +apiVersion: apps/v1 +kind: DaemonSet +metadata: + name: nvidia-driver-compat + namespace: cozy-gpu-operator + labels: + app: nvidia-driver-compat +spec: + selector: + matchLabels: + app: nvidia-driver-compat + template: + metadata: + labels: + app: nvidia-driver-compat + # DaemonSet rather than a one-shot Job: staging is idempotent per-node, + # must re-run after every Talos reboot (hostPath contents are wiped on + # reboot when the system extension re-installs), and the pause container + # keeps the pod visible for operator observability (kubectl get pods). + spec: + priorityClassName: system-node-critical + # Restrict to GPU nodes only. Without this the DaemonSet schedules onto + # every node (including control-plane and CPU-only workers) and burns + # resources on hosts where the compat tree is meaningless. + # The label is published by Node Feature Discovery / GPU Operator's NFD + # subchart; if NFD is disabled, replace this with whatever label your + # cluster uses to mark GPU hosts. + nodeSelector: + nvidia.com/gpu.present: "true" + # Tolerate-all is intentionally broad: the nodeSelector above already + # confines scheduling to GPU nodes, so the "Exists" toleration only + # matters when those nodes carry extra taints (dedicated=gpu:NoSchedule, + # nvidia.com/gpu:NoSchedule from the GPU Operator's default policy, + # etc.). This mirrors the stance of the upstream nvidia-driver-daemonset + # and nvidia-device-plugin DaemonSets — blanket tolerations plus a + # narrow nodeSelector. + tolerations: + - operator: Exists + initContainers: + - name: create-driver-tree + image: busybox:1.37 + command: + - sh + - -c + - | + set -e + GLIBC_LIB="/host/usr/local/glibc/usr/lib" + SRC_BIN="/host/usr/local/bin" + DST="/host/var/nvidia-driver" + + mkdir -p "$DST/bin" + + [ -f "$GLIBC_LIB/libnvidia-ml.so.1" ] || { + echo "missing $GLIBC_LIB/libnvidia-ml.so.1" >&2 + exit 1 + } + [ -f "$SRC_BIN/nvidia-smi" ] || { + echo "missing $SRC_BIN/nvidia-smi" >&2 + exit 1 + } + + cp -f "$GLIBC_LIB/libnvidia-ml.so.1" "$DST/libnvidia-ml.so.1" + echo "Copied libnvidia-ml.so.1" + + cp -f "$SRC_BIN/nvidia-smi" "$DST/bin/nvidia-smi" + chmod +x "$DST/bin/nvidia-smi" + echo "Copied nvidia-smi" + + mkdir -p /host/run/nvidia/validations + touch /host/run/nvidia/validations/.driver-ctr-ready + echo "Created driver-ctr-ready flag" + echo "Done" + securityContext: + privileged: true + resources: + requests: + cpu: 10m + memory: 16Mi + limits: + cpu: 100m + memory: 64Mi + volumeMounts: + - name: host-root + mountPath: /host + containers: + - name: pause + image: registry.k8s.io/pause:3.10 + resources: + requests: + cpu: 10m + memory: 8Mi + limits: + cpu: 50m + memory: 16Mi + volumes: + - name: host-root + hostPath: + path: / diff --git a/packages/system/gpu-operator/examples/values-native-talos.yaml b/packages/system/gpu-operator/examples/values-native-talos.yaml new file mode 100644 index 00000000..ef7b2891 --- /dev/null +++ b/packages/system/gpu-operator/examples/values-native-talos.yaml @@ -0,0 +1,52 @@ +# Cozystack Package values for running GPU workloads natively in pods +# on Talos. This is the counterpart to values-talos.yaml, which targets +# the sandbox (VFIO passthrough) scenario. +# +# Prerequisites: +# - NVIDIA Talos system extension installed on GPU nodes. +# - examples/nvidia-driver-compat.yaml deployed to stage driver files +# where the gpu-operator validator looks for them. +# - examples/dcgm-custom-metrics.yaml applied so DCGM Exporter exports +# the full set of metrics used by the dashboard and recording rules. +apiVersion: cozystack.io/v1alpha1 +kind: Package +metadata: + name: cozystack.gpu-operator +spec: + variant: default + components: + gpu-operator: + values: + gpu-operator: + # The compat DaemonSet stages libnvidia-ml.so.1 and nvidia-smi + # under /var/nvidia-driver. Point the validator at that path. + hostPaths: + driverInstallDir: "/var/nvidia-driver" + # Disable the sandbox path — workloads run in pods, not VMs. + sandboxWorkloads: + enabled: false + vfioManager: + enabled: false + # The Talos extension provides the driver and runtime hooks, + # so the operator's own toolkit and driver components must be + # switched off to avoid conflicting installations. + toolkit: + enabled: false + devicePlugin: + enabled: true + # Export full set of DCGM metrics using the ConfigMap in + # examples/dcgm-custom-metrics.yaml. + dcgmExporter: + serviceMonitor: + enabled: true + # Matches the 1m evaluation cadence of GPU recording rules with + # 4x oversampling headroom. DCGM_FI_PROF_* counters have + # documented overhead concerns below 1s; 15s is safe. + interval: "15s" + honorLabels: true + relabelings: + - sourceLabels: [__meta_kubernetes_pod_node_name] + targetLabel: node + action: replace + config: + name: dcgm-custom-metrics diff --git a/packages/system/hami/Chart.yaml b/packages/system/hami/Chart.yaml new file mode 100644 index 00000000..3fcf4c5d --- /dev/null +++ b/packages/system/hami/Chart.yaml @@ -0,0 +1,3 @@ +apiVersion: v2 +name: cozy-hami +version: 0.0.0 # Placeholder, the actual version will be automatically set during the build process diff --git a/packages/system/hami/Makefile b/packages/system/hami/Makefile new file mode 100644 index 00000000..83663a66 --- /dev/null +++ b/packages/system/hami/Makefile @@ -0,0 +1,43 @@ +export NAME=hami +export NAMESPACE=cozy-$(NAME) + +include ../../../hack/common-envs.mk +include ../../../hack/package.mk + +# When bumping the HAMi version, run `make update` and then review +# the resulting diff in `charts/hami/`. The recipe below reproduces the +# top-level vendoring overrides automatically: +# +# 1. Removes the broken hami-dra subchart. Upstream's NVIDIA DRA driver +# path requires kubelet DRA support that cozystack does not enable +# and has no upstream fix tracked. See commit 3c5521e. +# 2. Empties Chart.yaml dependencies and drops Chart.lock so Helm does +# not try to re-pull hami-dra at build time. See commit 2734dc0. +# 3. Strips dra/hami-dra/podSecurityPolicy blocks from the upstream +# values.yaml since the corresponding code paths are gone. PSP is +# removed from Kubernetes 1.25+ and is unused by cozystack. +# +# Template-level patches are NOT reproduced automatically: +# +# * Scheduler templates have `{{- if .Values.dra.enabled }}` blocks +# that need to be removed because the dra value is gone (commit +# 2734dc0 stripped them). +# * device-plugin/monitorservice.yaml uses `indent` with leading +# whitespace; it must be rewritten to `nindent` for the labels block +# to render correctly when devicePlugin.service.labels is set +# (commit 3685254). +# +# After `make update`, run `git diff -- charts/hami/templates/` and +# replay these template patches against the new upstream version, then +# verify with `helm unittest`. If upstream restructured the affected +# files, the patches may need to be redesigned rather than reapplied. + +update: + rm -rf charts + helm repo add hami-charts https://project-hami.github.io/HAMi/ + helm repo update hami-charts + helm pull hami-charts/hami --untar --untardir charts + rm -rf charts/hami/charts/hami-dra + yq --inplace '.dependencies = []' charts/hami/Chart.yaml + rm -f charts/hami/Chart.lock + yq --inplace 'del(.dra) | del(.["hami-dra"]) | del(.podSecurityPolicy)' charts/hami/values.yaml diff --git a/packages/system/hami/README.md b/packages/system/hami/README.md new file mode 100644 index 00000000..68669ac7 --- /dev/null +++ b/packages/system/hami/README.md @@ -0,0 +1,82 @@ +# HAMi — GPU Virtualization Middleware + +[HAMi](https://github.com/Project-HAMi/HAMi) (Heterogeneous AI Computing Virtualization Middleware) is a CNCF Sandbox project that enables fractional GPU sharing in Kubernetes. It allows workloads to request specific amounts of GPU memory and compute cores instead of claiming entire GPUs. + +## Architecture + +HAMi consists of four components: + +- **MutatingWebhook** — intercepts pod creation, injects `schedulerName: hami-scheduler` +- **Scheduler Extender** — extends kube-scheduler with GPU-aware Filter and Bind logic +- **Device Plugin** (DaemonSet) — registers vGPU resources via the Kubernetes Device Plugin API +- **HAMi-core** (`libvgpu.so`) — `LD_PRELOAD` library injected into workload containers, intercepts CUDA API calls to enforce memory and compute isolation + +## Prerequisites + +- GPU Operator must be enabled (`addons.gpuOperator.enabled: true`) +- NVIDIA driver >= 440 on host nodes +- nvidia-container-toolkit configured as the default container runtime +- GPU nodes labeled with `gpu=on` + +## Known Limitations + +### glibc < 2.34 requirement for workload containers + +HAMi-core uses `LD_PRELOAD` to intercept `dlsym()` for CUDA symbol resolution. The fallback code path relies on `_dl_sym`, a private glibc internal symbol that was removed in glibc 2.34 when libdl and libpthread were merged into libc.so. + +**This limitation affects workload containers only**, not the host OS or HAMi's own components. + +| Distribution | glibc | Result | +| --------------- | ----- | -------------------------------------------- | +| Ubuntu 18.04 | 2.27 | Full isolation (memory + compute) | +| Ubuntu 20.04 | 2.31 | Full isolation (memory + compute) | +| Ubuntu 22.04 | 2.35 | Memory isolation works, compute breaks | +| Ubuntu 24.04 | 2.39 | Both memory and compute isolation break | +| Alpine (musl) | N/A | Completely incompatible (`dlvsym` absent) | + +Most modern ML/AI base images (CUDA 12.x, PyTorch 2.x, TensorFlow 2.x) use Ubuntu 22.04+ with glibc >= 2.35, which means compute isolation will not work with these images until the upstream fix is merged. + +**Upstream tracking issues:** + +- [HAMi-core#174](https://github.com/Project-HAMi/HAMi-core/issues/174) — `_dl_sym` removal in glibc 2.34 breaks HAMi-core's CUDA symbol resolution at the symbol level +- [HAMi#1190](https://github.com/Project-HAMi/HAMi/issues/1190) — maintainer thread confirming the empirical per-glibc-version isolation behavior shown in the table above + +### musl libc (Alpine) incompatibility + +HAMi-core is completely incompatible with musl libc. The `dlvsym()` function used by HAMi-core is a glibc extension not available in musl. Only glibc-based container images (Debian, Ubuntu, RHEL, etc.) can use HAMi GPU isolation. + +## Usage + +Enable HAMi in your tenant Kubernetes cluster values: + +```yaml +addons: + gpuOperator: + enabled: true + hami: + enabled: true +``` + +When HAMi is enabled, GPU Operator's built-in device plugin is automatically disabled to avoid conflicts. This default is preserved by setting `addons.gpuOperator.valuesOverride.gpu-operator.devicePlugin.enabled: false`; advanced topologies that partition GPU pools (e.g. some nodes use HAMi while others run the standard NVIDIA device plugin via node selectors) can re-enable it explicitly through `valuesOverride`. + +### Requesting fractional GPU resources + +```yaml +resources: + limits: + nvidia.com/gpu: 1 + nvidia.com/gpumem: 3000 # 3000 MB of GPU memory + nvidia.com/gpucores: 30 # 30% of GPU compute cores +``` + +## Parameters + +Default values shown below are inherited from the upstream HAMi chart and may change with upstream updates. + +| Name | Description | Default | +| --- | --- | --- | +| `hami.devicePlugin.runtimeClassName` | RuntimeClass for device plugin pods | `nvidia` | +| `hami.devicePlugin.deviceSplitCount` | Max virtual GPUs per physical GPU | `10` | +| `hami.devicePlugin.deviceMemoryScaling` | Memory overcommit factor (> 1.0 enables overcommit) | `1` | +| `hami.scheduler.defaultSchedulerPolicy.nodeSchedulerPolicy` | Node packing strategy (`binpack` or `spread`) | `binpack` | +| `hami.scheduler.defaultSchedulerPolicy.gpuSchedulerPolicy` | GPU packing strategy (`binpack` or `spread`) | `spread` | diff --git a/packages/system/hami/charts/hami/Chart.yaml b/packages/system/hami/charts/hami/Chart.yaml new file mode 100644 index 00000000..55f32ab6 --- /dev/null +++ b/packages/system/hami/charts/hami/Chart.yaml @@ -0,0 +1,18 @@ +apiVersion: v2 +appVersion: 2.8.1 +dependencies: [] +description: Heterogeneous AI Computing Virtualization Middleware +keywords: +- vgpu +- gpu +kubeVersion: '>= 1.18.0-0' +maintainers: +- email: archlitchi@gmail.com + name: limengxuan +- email: xiaozhang0210@hotmail.com + name: zhangxiao +name: hami +sources: +- https://github.com/Project-HAMi/HAMi +type: application +version: 2.8.1 diff --git a/packages/system/hami/charts/hami/README.md b/packages/system/hami/charts/hami/README.md new file mode 100644 index 00000000..e8217290 --- /dev/null +++ b/packages/system/hami/charts/hami/README.md @@ -0,0 +1,237 @@ +# HAMi Helm Chart Values Documentation + +This document provides detailed descriptions of all configurable values parameters for the HAMi Helm Chart. + +## Global Configuration + +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `global.imageRegistry` | Global Docker image registry | `""` | +| `global.imagePullSecrets` | Global Docker image pull secrets | `[]` | +| `global.imageTag` | Image tag | `"v2.8.1"` | +| `global.gpuHookPath` | GPU Hook path | `/usr/local` | +| `global.labels` | Global labels | `{}` | +| `global.annotations` | Global annotations | `{}` | +| `global.managedNodeSelectorEnable` | Whether to enable managed node selector | `false` | +| `global.managedNodeSelector.usage` | Managed node selector usage | `"gpu"` | +| `nameOverride` | Name override | `""` | +| `fullnameOverride` | Full name override | `""` | +| `namespaceOverride` | Namespace override | `""` | + +## Resource Name Configuration + +### NVIDIA GPU Resources +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `resourceName` | GPU resource name | `"nvidia.com/gpu"` | +| `resourceMem` | GPU memory resource name | `"nvidia.com/gpumem"` | +| `resourceMemPercentage` | GPU memory percentage resource name | `"nvidia.com/gpumem-percentage"` | +| `resourceCores` | GPU core resource name | `"nvidia.com/gpucores"` | +| `resourcePriority` | GPU priority resource name | `"nvidia.com/priority"` | + +### Cambricon MLU Resources +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `mluResourceName` | MLU resource name | `"cambricon.com/vmlu"` | +| `mluResourceMem` | MLU memory resource name | `"cambricon.com/mlu.smlu.vmemory"` | +| `mluResourceCores` | MLU core resource name | `"cambricon.com/mlu.smlu.vcore"` | + +### Hygon DCU Resources +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `dcuResourceName` | DCU resource name | `"hygon.com/dcunum"` | +| `dcuResourceMem` | DCU memory resource name | `"hygon.com/dcumem"` | +| `dcuResourceCores` | DCU core resource name | `"hygon.com/dcucores"` | + +### Metax GPU Resources +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `metaxResourceName` | GPU resource name | `"metax-tech.com/sgpu"` | +| `metaxResourceCore` | GPU core resource name | `"metax-tech.com/vcore"` | +| `metaxResourceMem` | GPU memory resource name | `"metax-tech.com/vmemory"` | +| `metaxsGPUTopologyAware` | GPU topology awareness | `"false"` | + +### Enflame GCU Resources +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `enflameResourceNameVGCU` | vGCU resource name | `"enflame.com/vgcu"` | +| `enflameResourceNameVGCUPercentage` | vGCU percentage resource name | `"enflame.com/vgcu-percentage"` | + +### Kunlunxin XPU Resources +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `kunlunResourceName` | XPU resource name | `"kunlunxin.com/xpu"` | + +## Scheduler Configuration + +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `schedulerName` | Scheduler name | `"hami-scheduler"` | +| `scheduler.nodeName` | Define node name, scheduler will schedule to this node | `""` | +| `scheduler.overwriteEnv` | Whether to overwrite environment variables | `"false"` | +| `scheduler.defaultSchedulerPolicy.nodeSchedulerPolicy` | Node scheduler policy | `binpack` | +| `scheduler.defaultSchedulerPolicy.gpuSchedulerPolicy` | GPU scheduler policy | `spread` | +| `scheduler.metricsBindAddress` | Metrics bind address | `":9395"` | +| `scheduler.forceOverwriteDefaultScheduler` | Whether to force overwrite default scheduler | `true` | +| `scheduler.livenessProbe` | Whether to enable liveness probe | `false` | +| `scheduler.leaderElect` | Whether to enable leader election | `true` | +| `scheduler.replicas` | Number of replicas | `1` | + +### Kube Scheduler Configuration + +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `scheduler.kubeScheduler.enabled` | Whether to run kube-scheduler container in scheduler pod | `true` | +| `scheduler.kubeScheduler.image.registry` | Kube scheduler image registry | `"registry.cn-hangzhou.aliyuncs.com"` | +| `scheduler.kubeScheduler.image.repository` | Kube scheduler image repository | `"google_containers/kube-scheduler"` | +| `scheduler.kubeScheduler.image.tag` | Kube scheduler image tag | `""` | +| `scheduler.kubeScheduler.image.pullPolicy` | Kube scheduler image pull policy | `IfNotPresent` | +| `scheduler.kubeScheduler.image.pullSecrets` | Kube scheduler image pull secrets | `[]` | +| `scheduler.kubeScheduler.extraNewArgs` | Extra new arguments | `["--config=/config/config.yaml", "-v=4"]` | +| `scheduler.kubeScheduler.extraArgs` | Extra arguments | `["--policy-config-file=/config/config.json", "-v=4"]` | + +### Extender Configuration + +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `scheduler.extender.image.registry` | Scheduler extender image registry | `"docker.io"` | +| `scheduler.extender.image.repository` | Scheduler extender image repository | `"projecthami/hami"` | +| `scheduler.extender.image.tag` | Scheduler extender image tag | `""` | +| `scheduler.extender.image.pullPolicy` | Scheduler extender image pull policy | `IfNotPresent` | +| `scheduler.extender.image.pullSecrets` | Scheduler extender image pull secrets | `[]` | +| `scheduler.extender.extraArgs` | Scheduler extender extra arguments | `["--debug", "-v=4"]` | + +### Admission Webhook Configuration + +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `scheduler.admissionWebhook.enabled` | Whether to enable admission webhook | `true` | +| `scheduler.admissionWebhook.customURL.enabled` | Whether to enable custom URL | `false` | +| `scheduler.admissionWebhook.customURL.host` | Custom URL host | `127.0.0.1` | +| `scheduler.admissionWebhook.customURL.port` | Custom URL port | `31998` | +| `scheduler.admissionWebhook.customURL.path` | Custom URL path | `/webhook` | +| `scheduler.admissionWebhook.reinvocationPolicy` | Reinvocation policy | `Never` | +| `scheduler.admissionWebhook.failurePolicy` | Failure policy | `Ignore` | + +### TLS Certificate Configuration + +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `scheduler.certManager.enabled` | Whether to use cert-manager to generate self-signed certificates | `false` | +| `scheduler.patch.enabled` | Whether to use kube-webhook-certgen to generate self-signed certificates | `true` | +| `scheduler.patch.image.registry` | Certgen image registry | `"docker.io"` | +| `scheduler.patch.image.repository` | Certgen image repository | `"jettech/kube-webhook-certgen"` | +| `scheduler.patch.image.tag` | Certgen image tag | `"v1.5.2"` | +| `scheduler.patch.image.pullPolicy` | Certgen image pull policy | `IfNotPresent` | +| `scheduler.patch.imageNew.registry` | New certgen image registry | `"docker.io"` | +| `scheduler.patch.imageNew.repository` | New certgen image repository | `"liangjw/kube-webhook-certgen"` | +| `scheduler.patch.imageNew.tag` | New certgen image tag | `"v1.1.1"` | + +### Scheduler Service Configuration + +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `scheduler.service.type` | Service type | `NodePort` | +| `scheduler.service.httpPort` | HTTP port | `443` | +| `scheduler.service.schedulerPort` | Scheduler NodePort | `31998` | +| `scheduler.service.monitorPort` | Monitor port | `31993` | +| `scheduler.service.monitorTargetPort` | Monitor target port | `9395` | + +## Device Plugin Configuration + +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `devicePlugin.image.registry` | Device plugin image registry | `"docker.io"` | +| `devicePlugin.image.repository` | Device plugin image repository | `"projecthami/hami"` | +| `devicePlugin.image.tag` | Device plugin image tag | `""` | +| `devicePlugin.image.pullPolicy` | Device plugin image pull policy | `IfNotPresent` | +| `devicePlugin.image.pullSecrets` | Device plugin image pull secrets | `[]` | + +### Monitor Configuration + +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `devicePlugin.monitor.image.registry` | Monitor image registry | `"docker.io"` | +| `devicePlugin.monitor.image.repository` | Monitor image repository | `"projecthami/hami"` | +| `devicePlugin.monitor.image.tag` | Monitor image tag | `""` | +| `devicePlugin.monitor.image.pullPolicy` | Monitor image pull policy | `IfNotPresent` | +| `devicePlugin.monitor.image.pullSecrets` | Monitor image pull secrets | `[]` | +| `devicePlugin.monitor.ctrPath` | Container path | `/usr/local/vgpu/containers` | +| `devicePlugin.monitor.extraArgs` | Monitor extra arguments | `["-v=4"]` | +| `devicePlugin.monitor.extraEnvs` | Monitor extra environments | `{}` | + +### Device Plugin Other Configuration + +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `devicePlugin.deviceSplitCount` | Integer type, default value: 10. Maximum number of tasks assigned to a single GPU device | `10` | +| `devicePlugin.deviceMemoryScaling` | Device memory scaling ratio | `1` | +| `devicePlugin.deviceCoreScaling` | Device core scaling ratio | `1` | +| `devicePlugin.runtimeClassName` | Runtime class name | `""` | +| `devicePlugin.createRuntimeClass` | Whether to create runtime class | `false` | +| `devicePlugin.migStrategy` | String type, "none" means ignore MIG functionality, "mixed" means allocate MIG devices through independent resources | `"none"` | +| `devicePlugin.disablecorelimit` | String type, "true" means disable core limit, "false" means enable core limit | `"false"` | +| `devicePlugin.passDeviceSpecsEnabled` | Whether to enable passing device specs | `false` | +| `devicePlugin.extraArgs` | Device plugin extra arguments | `["-v=4"]` | +| `devicePlugin.nodeConfiguration.config` | Node configuration for device plugin by json | An example of default configuration. | +| `devicePlugin.nodeConfiguration.externalConfigName` | Node configuration for device plugin by external congimap | `""` | +| `devicePlugin.extraEnvs` | Device plugin extra environments | `{}` | + +### Device Plugin Service Configuration + +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `devicePlugin.service.type` | Service type | `NodePort` | +| `devicePlugin.service.httpPort` | HTTP port | `31992` | + +### Device Plugin Deployment Configuration + +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `devicePlugin.pluginPath` | Plugin path | `/var/lib/kubelet/device-plugins` | +| `devicePlugin.libPath` | Library path | `/usr/local/vgpu` | +| `devicePlugin.nvidiaNodeSelector` | NVIDIA node selector | `{"gpu": "on"}` | +| `devicePlugin.updateStrategy.type` | Update strategy type | `RollingUpdate` | +| `devicePlugin.updateStrategy.rollingUpdate.maxUnavailable` | Maximum unavailable count | `1` | + +## Device Configuration + +### AWS Neuron +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `devices.awsneuron.customresources` | Custom resources | `["aws.amazon.com/neuron", "aws.amazon.com/neuroncore"]` | + +### Kunlunxin +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `devices.kunlun.enabled` | Whether to enable | `true` | +| `devices.kunlun.customresources` | Custom resources | `["kunlunxin.com/xpu"]` | + +### Mthreads +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `devices.mthreads.enabled` | Whether to enable | `true` | +| `devices.mthreads.customresources` | Custom resources | `["mthreads.com/vgpu"]` | + +### NVIDIA +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `devices.nvidia.gpuCorePolicy` | GPU core policy | `default` | +| `devices.nvidia.libCudaLogLevel` | CUDA library log level | `1` | + +### Huawei Ascend +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `devices.ascend.enabled` | Whether to enable | `false` | +| `devices.ascend.image` | Image | `""` | +| `devices.ascend.imagePullPolicy` | Image pull policy | `IfNotPresent` | +| `devices.ascend.extraArgs` | Extra arguments | `[]` | +| `devices.ascend.nodeSelector` | Node selector | `{"ascend": "on"}` | +| `devices.ascend.tolerations` | Tolerations | `[]` | +| `devices.ascend.customresources` | Custom resources | `["huawei.com/Ascend910A", "huawei.com/Ascend910A-memory", ...]` | + +### Iluvatar +| Parameter | Description | Default Value | +|-----------|-------------|---------------| +| `devices.iluvatar.enabled` | Whether to enable | `false` | +| `devices.iluvatar.customresources` | Custom resources | `["iluvatar.ai/BI-V150-vgpu", "iluvatar.ai/BI-V150.vMem","iluvatar.ai/BI-V150.vCore", ...]` | diff --git a/packages/system/hami/charts/hami/templates/NOTES.txt b/packages/system/hami/charts/hami/templates/NOTES.txt new file mode 100644 index 00000000..15bb1218 --- /dev/null +++ b/packages/system/hami/charts/hami/templates/NOTES.txt @@ -0,0 +1,3 @@ +** Please be patient while the chart is being deployed ** +Resource name: {{ .Values.resourceName }} + diff --git a/packages/system/hami/charts/hami/templates/_commons.tpl b/packages/system/hami/charts/hami/templates/_commons.tpl new file mode 100644 index 00000000..b68018e4 --- /dev/null +++ b/packages/system/hami/charts/hami/templates/_commons.tpl @@ -0,0 +1,49 @@ +{{/* +Return the proper image name +{{ include "common.images.image" ( dict "imageRoot" .Values.path.to.the.image "global" $) }} +*/}} +{{- define "common.images.image" -}} +{{- $registryName := .imageRoot.registry -}} +{{- $repositoryName := .imageRoot.repository -}} +{{- $tag := .imageRoot.tag | toString -}} +{{- if .global }} + {{- if .global.imageRegistry }} + {{- $registryName = .global.imageRegistry -}} + {{- end -}} +{{- end -}} +{{- if .tag }} + {{- $tag = .tag | toString -}} +{{- end -}} +{{- if $registryName }} +{{- printf "%s/%s:%s" $registryName $repositoryName $tag -}} +{{- else -}} +{{- printf "%s:%s" $repositoryName $tag -}} +{{- end -}} +{{- end -}} + +{{/* +Return the proper Docker Image Registry Secret Names (deprecated: use common.images.renderPullSecrets instead) +{{ include "common.images.pullSecrets" ( dict "images" (list .Values.path.to.the.image1, .Values.path.to.the.image2) "global" .Values.global) }} +*/}} +{{- define "common.images.pullSecrets" -}} + {{- $pullSecrets := list }} + + {{- if .global }} + {{- range .global.imagePullSecrets -}} + {{- $pullSecrets = append $pullSecrets . -}} + {{- end -}} + {{- end -}} + + {{- range .images -}} + {{- range .pullSecrets -}} + {{- $pullSecrets = append $pullSecrets . -}} + {{- end -}} + {{- end -}} + + {{- if (not (empty $pullSecrets)) }} +imagePullSecrets: + {{- range $pullSecrets }} + - name: {{ . }} + {{- end }} + {{- end }} +{{- end -}} diff --git a/packages/system/hami/charts/hami/templates/_helpers.tpl b/packages/system/hami/charts/hami/templates/_helpers.tpl new file mode 100644 index 00000000..ffcc61da --- /dev/null +++ b/packages/system/hami/charts/hami/templates/_helpers.tpl @@ -0,0 +1,163 @@ +{{/* +Expand the name of the chart. +*/}} +{{- define "hami-vgpu.name" -}} +{{- default .Chart.Name .Values.nameOverride | trunc 63 | trimSuffix "-" -}} +{{- end -}} + +{{/* +Create a default fully qualified app name. +We truncate at 63 chars because some Kubernetes name fields are limited to this (by the DNS naming spec). +If release name contains chart name it will be used as a full name. +*/}} +{{- define "hami-vgpu.fullname" -}} +{{- if .Values.fullnameOverride -}} +{{- .Values.fullnameOverride | trunc 63 | trimSuffix "-" -}} +{{- else }} +{{- $name := default .Chart.Name .Values.nameOverride -}} +{{- if contains $name .Release.Name }} +{{- .Release.Name | trunc 63 | trimSuffix "-" -}} +{{- else -}} +{{- printf "%s-%s" .Release.Name $name | trunc 63 | trimSuffix "-" -}} +{{- end -}} +{{- end -}} +{{- end -}} + +{{/* +Allow the release namespace to be overridden for multi-namespace deployments in combined charts +*/}} +{{- define "hami-vgpu.namespace" -}} + {{- if .Values.namespaceOverride -}} + {{- .Values.namespaceOverride -}} + {{- else -}} + {{- .Release.Namespace -}} + {{- end -}} +{{- end -}} + +{{/* +The app name for Scheduler +*/}} +{{- define "hami-vgpu.scheduler" -}} +{{- printf "%s-scheduler" ( include "hami-vgpu.fullname" . ) | trunc 63 | trimSuffix "-" -}} +{{- end -}} + +{{/* +The app name for DevicePlugin +*/}} +{{- define "hami-vgpu.device-plugin" -}} +{{- printf "%s-device-plugin" ( include "hami-vgpu.fullname" . ) | trunc 63 | trimSuffix "-" -}} +{{- end -}} + +{{/* + The app name for MockDevicePlugin + */}} +{{- define "hami-vgpu.mock-device-plugin" -}} +{{- printf "%s-mock-device-plugin" ( include "hami-vgpu.fullname" . ) | trunc 63 | trimSuffix "-" -}} +{{- end -}} + +{{/* +The tls secret name for Scheduler +*/}} +{{- define "hami-vgpu.scheduler.tls" -}} +{{- printf "%s-scheduler-tls" ( include "hami-vgpu.fullname" . ) | trunc 63 | trimSuffix "-" -}} +{{- end -}} + +{{/* +The webhook name +*/}} +{{- define "hami-vgpu.scheduler.webhook" -}} +{{- printf "%s-webhook" ( include "hami-vgpu.fullname" . ) | trunc 63 | trimSuffix "-" -}} +{{- end -}} + +{{/* +Create chart name and version as used by the chart label. +*/}} +{{- define "hami-vgpu.chart" -}} +{{- printf "%s-%s" .Chart.Name .Chart.Version | replace "+" "_" | trunc 63 | trimSuffix "-" }} +{{- end }} + +{{/* +Common labels +*/}} +{{- define "hami-vgpu.labels" -}} +helm.sh/chart: {{ include "hami-vgpu.chart" . }} +{{ include "hami-vgpu.selectorLabels" . }} +{{- if .Chart.AppVersion }} +app.kubernetes.io/version: {{ .Chart.AppVersion | quote }} +{{- end }} +app.kubernetes.io/managed-by: {{ .Release.Service }} +{{- end }} + +{{/* +Selector labels +*/}} +{{- define "hami-vgpu.selectorLabels" -}} +app.kubernetes.io/name: {{ include "hami-vgpu.name" . }} +app.kubernetes.io/instance: {{ .Release.Name }} +{{- end }} + + +{{/* + Resolve the tag for kubeScheduler. +*/}} +{{- define "resolvedKubeSchedulerTag" -}} +{{- if .Values.scheduler.kubeScheduler.image.tag }} +{{- .Values.scheduler.kubeScheduler.image.tag | trim -}} +{{- else }} +{{- include "strippedKubeVersion" . | trim -}} +{{- end }} +{{- end }} + +{{/* + Return the stripped Kubernetes version string by removing extra parts after semantic version number. + v1.31.1+k3s1 -> v1.31.1 + v1.30.8-eks-2d5f260 -> v1.30.8 + v1.31.1 -> v1.31.1 +*/}} +{{- define "strippedKubeVersion" -}} +{{ regexReplaceAll "^(v[0-9]+\\.[0-9]+\\.[0-9]+)(.*)$" .Capabilities.KubeVersion.Version "$1" }} +{{- end -}} + +{{- define "hami.scheduler.kubeScheduler.image" -}} +{{ include "common.images.image" (dict "imageRoot" .Values.scheduler.kubeScheduler.image "global" .Values.global "tag" (include "resolvedKubeSchedulerTag" .)) }} +{{- end -}} + +{{- define "hami.scheduler.extender.image" -}} +{{ include "common.images.image" (dict "imageRoot" .Values.scheduler.extender.image "global" .Values.global "tag" .Values.global.imageTag) }} +{{- end -}} + +{{- define "hami.devicePlugin.image" -}} +{{ include "common.images.image" (dict "imageRoot" .Values.devicePlugin.image "global" .Values.global "tag" .Values.global.imageTag) }} +{{- end -}} + +{{- define "hami.mockDevicePlugin.image" -}} +{{ include "common.images.image" (dict "imageRoot" .Values.mockDevicePlugin.image "global" .Values.global "tag" .Values.mockDevicePlugin.tag) }} +{{- end -}} + +{{- define "hami.devicePlugin.monitor.image" -}} +{{ include "common.images.image" (dict "imageRoot" .Values.devicePlugin.monitor.image "global" .Values.global "tag" .Values.global.imageTag) }} +{{- end -}} + +{{- define "hami.scheduler.patch.image" -}} +{{ include "common.images.image" (dict "imageRoot" .Values.scheduler.patch.image "global" .Values.global) }} +{{- end -}} + +{{- define "hami.scheduler.patch.new.image" -}} +{{ include "common.images.image" (dict "imageRoot" .Values.scheduler.patch.imageNew "global" .Values.global) }} +{{- end -}} + +{{- define "hami.scheduler.extender.imagePullSecrets" -}} +{{ include "common.images.pullSecrets" (dict "images" (list .Values.scheduler.extender.image) "global" .Values.global) }} +{{- end -}} + +{{- define "hami.devicePlugin.imagePullSecrets" -}} +{{ include "common.images.pullSecrets" (dict "images" (list .Values.devicePlugin.image) "global" .Values.global) }} +{{- end -}} + +{{- define "hami.scheduler.patch.imagePullSecrets" -}} +{{ include "common.images.pullSecrets" (dict "images" (list .Values.scheduler.patch.image) "global" .Values.global) }} +{{- end -}} + +{{- define "hami.scheduler.patch.new.imagePullSecrets" -}} +{{ include "common.images.pullSecrets" (dict "images" (list .Values.scheduler.patch.imageNew) "global" .Values.global) }} +{{- end -}} diff --git a/packages/system/hami/charts/hami/templates/device-plugin/configmap.yaml b/packages/system/hami/charts/hami/templates/device-plugin/configmap.yaml new file mode 100644 index 00000000..74631e24 --- /dev/null +++ b/packages/system/hami/charts/hami/templates/device-plugin/configmap.yaml @@ -0,0 +1,13 @@ +{{- if and .Values.devicePlugin.enabled (not .Values.devicePlugin.nodeConfiguration.externalConfigName) -}} +apiVersion: v1 +kind: ConfigMap +metadata: + name: {{ include "hami-vgpu.device-plugin" . }} + namespace: {{ include "hami-vgpu.namespace" . }} + labels: + app.kubernetes.io/component: hami-device-plugin + {{- include "hami-vgpu.labels" . | nindent 4 }} +data: + config.json: | +{{- .Values.devicePlugin.nodeConfiguration.config | nindent 4 }} +{{- end }} diff --git a/packages/system/hami/charts/hami/templates/device-plugin/daemonsetmock.yaml b/packages/system/hami/charts/hami/templates/device-plugin/daemonsetmock.yaml new file mode 100644 index 00000000..a0dc4ff3 --- /dev/null +++ b/packages/system/hami/charts/hami/templates/device-plugin/daemonsetmock.yaml @@ -0,0 +1,55 @@ +{{- if .Values.mockDevicePlugin.enabled }} +apiVersion: apps/v1 +kind: DaemonSet +metadata: + name: {{ include "hami-vgpu.mock-device-plugin" . }} + namespace: {{ include "hami-vgpu.namespace" . }} +spec: + selector: + matchLabels: + app.kubernetes.io/component: hami-mock-device-plugin + {{- include "hami-vgpu.selectorLabels" . | nindent 6 }} + template: + metadata: + annotations: + scheduler.alpha.kubernetes.io/critical-pod: "" + labels: + app.kubernetes.io/component: hami-mock-device-plugin + {{- include "hami-vgpu.selectorLabels" . | nindent 8 }} + spec: + serviceAccountName: {{ include "hami-vgpu.mock-device-plugin" . }} + tolerations: + - key: CriticalAddonsOnly + operator: Exists + containers: + - image: {{ include "hami.mockDevicePlugin.image" . }} + imagePullPolicy: {{ .Values.mockDevicePlugin.image.pullPolicy }} + name: hami-mock-dp-cntr + env: + - name: NODE_NAME + valueFrom: + fieldRef: + fieldPath: spec.nodeName + command: + - ./k8s-device-plugin + - -v=5 + - --device-config-file=/device-config.yaml + volumeMounts: + - name: dp + mountPath: /var/lib/kubelet/device-plugins + - name: sys + mountPath: /sys + - name: device-config + mountPath: /device-config.yaml + subPath: device-config.yaml + volumes: + - name: dp + hostPath: + path: /var/lib/kubelet/device-plugins + - name: sys + hostPath: + path: /sys + - name: device-config + configMap: + name: {{ include "hami-vgpu.scheduler" . }}-device +{{- end -}} diff --git a/packages/system/hami/charts/hami/templates/device-plugin/daemonsetnvidia.yaml b/packages/system/hami/charts/hami/templates/device-plugin/daemonsetnvidia.yaml new file mode 100644 index 00000000..1f4c24f3 --- /dev/null +++ b/packages/system/hami/charts/hami/templates/device-plugin/daemonsetnvidia.yaml @@ -0,0 +1,262 @@ +{{- if .Values.devicePlugin.enabled }} +apiVersion: apps/v1 +kind: DaemonSet +metadata: + name: {{ include "hami-vgpu.device-plugin" . }} + namespace: {{ include "hami-vgpu.namespace" . }} + labels: + app.kubernetes.io/component: hami-device-plugin + {{- include "hami-vgpu.labels" . | nindent 4 }} + {{- with .Values.global.labels }} + {{- toYaml . | nindent 4 }} + {{- end }} + {{- if .Values.global.annotations }} + annotations: {{ toYaml .Values.global.annotations | nindent 4}} + {{- end }} +spec: + updateStrategy: + {{- with .Values.devicePlugin.updateStrategy }} + {{- toYaml . | nindent 4 }} + {{- end }} + selector: + matchLabels: + app.kubernetes.io/component: hami-device-plugin + {{- include "hami-vgpu.selectorLabels" . | nindent 6 }} + template: + metadata: + labels: + app.kubernetes.io/component: hami-device-plugin + hami.io/webhook: ignore + {{- include "hami-vgpu.selectorLabels" . | nindent 8 }} + annotations: + {{- if ge (regexReplaceAll "[^0-9]" .Capabilities.KubeVersion.Minor "" | int) 22 }} + checksum/hami-scheduler-newversion-config: {{ include (print $.Template.BasePath "/scheduler/configmapnew.yaml") . | sha256sum }} + {{- else }} + checksum/hami-scheduler-config: {{ include (print $.Template.BasePath "/scheduler/configmap.yaml") . | sha256sum }} + {{- end }} + checksum/hami-scheduler-device-config: {{ include (print $.Template.BasePath "/scheduler/device-configmap.yaml") . | sha256sum }} + {{- if .Values.devicePlugin.podAnnotations }} + {{- toYaml .Values.devicePlugin.podAnnotations | nindent 8 }} + {{- end }} + spec: + {{- if .Values.devicePlugin.runtimeClassName }} + runtimeClassName: {{ .Values.devicePlugin.runtimeClassName }} + {{- end }} + serviceAccountName: {{ include "hami-vgpu.device-plugin" . }} + priorityClassName: system-node-critical + hostPID: true + hostNetwork: true + {{- include "hami.devicePlugin.imagePullSecrets" . | nindent 6 }} + {{- if .Values.devicePlugin.gpuOperatorToolkitReady.enabled }} + initContainers: + - name: toolkit-validation + image: {{ include "hami.devicePlugin.image" . }} + imagePullPolicy: {{ .Values.devicePlugin.image.pullPolicy }} + securityContext: + privileged: true + runAsUser: 0 + command: ["sh", "-c"] + args: + - | + echo "Waiting for NVIDIA Toolkit to be ready..." + until [ -f {{ .Values.devicePlugin.gpuOperatorToolkitReady.hostPath }}/toolkit-ready ]; do + echo "Waiting for {{ .Values.devicePlugin.gpuOperatorToolkitReady.hostPath }}/toolkit-ready..." + sleep 5 + done + echo "NVIDIA Toolkit is ready!" + volumeMounts: + - name: nvidia-validations + mountPath: {{ .Values.devicePlugin.gpuOperatorToolkitReady.hostPath | quote }} + mountPropagation: HostToContainer + readOnly: true + {{- end }} + containers: + - name: device-plugin + image: {{ include "hami.devicePlugin.image" . }} + imagePullPolicy: {{ .Values.devicePlugin.image.pullPolicy }} + lifecycle: + postStart: + exec: + command: ["/bin/sh","-c", {{ printf "/k8s-vgpu/bin/vgpu-init.sh %s/vgpu/" .Values.global.gpuHookPath | quote }}] + command: + - nvidia-device-plugin + - --config-file=/device-config.yaml + - --mig-strategy={{ .Values.devicePlugin.migStrategy }} + - --disable-core-limit={{ .Values.devicePlugin.disablecorelimit }} + {{- range .Values.devicePlugin.extraArgs }} + - {{ . }} + {{- end }} + env: + - name: NODE_NAME + valueFrom: + fieldRef: + fieldPath: spec.nodeName + - name: NVIDIA_MIG_MONITOR_DEVICES + value: all + - name: DEVICE_LIST_STRATEGY + value: {{ .Values.devicePlugin.deviceListStrategy }} + - name: HOOK_PATH + value: {{ .Values.global.gpuHookPath }} + {{- if typeIs "bool" .Values.devicePlugin.passDeviceSpecsEnabled }} + - name: PASS_DEVICE_SPECS + value: {{ .Values.devicePlugin.passDeviceSpecsEnabled | quote }} + {{- end }} + {{- if typeIs "string" .Values.devicePlugin.nvidiaDriverRoot }} + - name: NVIDIA_DRIVER_ROOT + value: {{ .Values.devicePlugin.nvidiaDriverRoot }} + {{- end }} + {{- if typeIs "string" .Values.devicePlugin.nvidiaHookPath }} + - name: NVIDIA_CDI_HOOK_PATH + value: {{ .Values.devicePlugin.nvidiaHookPath }} + {{- end }} + {{- if typeIs "bool" .Values.devicePlugin.gdrcopyEnabled }} + - name: GDRCOPY_ENABLED + value: {{ .Values.devicePlugin.gdrcopyEnabled | quote }} + {{- end }} + {{- if typeIs "bool" .Values.devicePlugin.gdsEnabled }} + - name: GDS_ENABLED + value: {{ .Values.devicePlugin.gdsEnabled | quote }} + {{- end }} + {{- if typeIs "bool" .Values.devicePlugin.mofedEnabled }} + - name: MOFED_ENABLED + value: {{ .Values.devicePlugin.mofedEnabled | quote }} + {{- end }} + {{- if eq (.Values.scheduler.defaultSchedulerPolicy.gpuSchedulerPolicy | default "spread") "topology-aware" }} + - name: ENABLE_TOPOLOGY_SCORE + value: "true" + {{- end }} + {{- with .Values.devicePlugin.extraEnvs }} + {{- . | toYaml | nindent 12 }} + {{- end }} + securityContext: + privileged: true + allowPrivilegeEscalation: true + capabilities: + drop: ["ALL"] + add: ["SYS_ADMIN"] + resources: + {{- toYaml .Values.devicePlugin.resources | nindent 12 }} + volumeMounts: + - name: device-plugin + mountPath: /var/lib/kubelet/device-plugins + - name: lib + mountPath: {{ printf "%s%s" .Values.global.gpuHookPath "/vgpu" }} + - name: usrbin + mountPath: /usrbin + - name: deviceconfig + mountPath: /config + - name: hosttmp + mountPath: /tmp + - name: device-config + mountPath: /device-config.yaml + subPath: device-config.yaml + - name: cdi-root + mountPath: /var/run/cdi + {{- if typeIs "string" .Values.devicePlugin.nvidiaDriverRoot }} + # We always mount the driver root at /driver-root in the container. + # This is required for CDI detection to work correctly. + - name: driver-root + mountPath: /driver-root + readOnly: true + {{- end }} + - name: vgpu-monitor + image: {{ include "hami.devicePlugin.monitor.image" . }} + imagePullPolicy: {{ .Values.devicePlugin.monitor.image.pullPolicy }} + command: + - "vGPUmonitor" + {{- range .Values.devicePlugin.monitor.extraArgs }} + - {{ . }} + {{- end }} + securityContext: + allowPrivilegeEscalation: false + capabilities: + drop: ["ALL"] + add: ["SYS_ADMIN"] + env: + - name: NODE_NAME + valueFrom: + fieldRef: + fieldPath: spec.nodeName + - name: NVIDIA_VISIBLE_DEVICES + value: "all" + - name: NVIDIA_MIG_MONITOR_DEVICES + value: "all" + - name: HOOK_PATH + value: "{{ .Values.global.gpuHookPath }}/vgpu" + - name: HAMI_RESYNC_INTERVAL + value: {{ .Values.devicePlugin.monitor.resyncInterval | default "5m" | quote }} + {{- with .Values.devicePlugin.monitor.extraEnvs }} + {{- . | toYaml | nindent 12 }} + {{- end }} + resources: + {{- toYaml .Values.devicePlugin.monitor.resources | nindent 12 }} + volumeMounts: + - name: ctrs + mountPath: {{ .Values.devicePlugin.monitor.ctrPath }} + - name: dockers + mountPath: /run/docker + - name: containerds + mountPath: /run/containerd + - name: sysinfo + mountPath: /sysinfo + - name: hostvar + mountPath: /hostvar + - name: hosttmp + mountPath: /tmp + volumes: + - name: ctrs + hostPath: + path: {{ .Values.devicePlugin.monitor.ctrPath }} + - name: hosttmp + hostPath: + path: /tmp + - name: dockers + hostPath: + path: /run/docker + - name: containerds + hostPath: + path: /run/containerd + - name: device-plugin + hostPath: + path: {{ .Values.devicePlugin.pluginPath }} + - name: lib + hostPath: + path: {{ .Values.devicePlugin.libPath }} + {{- if .Values.devicePlugin.gpuOperatorToolkitReady.enabled }} + - name: nvidia-validations + hostPath: + path: {{ .Values.devicePlugin.gpuOperatorToolkitReady.hostPath }} + type: DirectoryOrCreate + {{- end }} + {{- if typeIs "string" .Values.devicePlugin.nvidiaDriverRoot }} + - name: driver-root + hostPath: + path: {{ .Values.devicePlugin.nvidiaDriverRoot }} + type: Directory + {{- end }} + - name: cdi-root + hostPath: + path: /var/run/cdi + type: DirectoryOrCreate + - name: usrbin + hostPath: + path: /usr/bin + - name: sysinfo + hostPath: + path: /sys + - name: hostvar + hostPath: + path: /var + - name: deviceconfig + configMap: + name: {{ .Values.devicePlugin.nodeConfiguration.externalConfigName | default (include "hami-vgpu.device-plugin" .) }} + - name: device-config + configMap: + name: {{ include "hami-vgpu.scheduler" . }}-device + {{- if .Values.devicePlugin.nvidiaNodeSelector }} + nodeSelector: {{ toYaml .Values.devicePlugin.nvidiaNodeSelector | nindent 8 }} + {{- end }} + {{- if .Values.devicePlugin.tolerations }} + tolerations: {{ toYaml .Values.devicePlugin.tolerations | nindent 8 }} + {{- end }} +{{- end }} diff --git a/packages/system/hami/charts/hami/templates/device-plugin/monitorrole.yaml b/packages/system/hami/charts/hami/templates/device-plugin/monitorrole.yaml new file mode 100644 index 00000000..6ac757f2 --- /dev/null +++ b/packages/system/hami/charts/hami/templates/device-plugin/monitorrole.yaml @@ -0,0 +1,28 @@ +{{- if .Values.devicePlugin.enabled -}} +apiVersion: rbac.authorization.k8s.io/v1 +kind: ClusterRole +metadata: + name: {{ include "hami-vgpu.device-plugin" . }}-monitor +rules: + - apiGroups: + - "" + resources: + - pods + verbs: + - get + - create + - watch + - list + - update + - patch + - apiGroups: + - "" + resources: + - nodes + verbs: + - get + - update + - list + - patch +{{- end -}} + diff --git a/packages/system/hami/charts/hami/templates/device-plugin/monitorrolebinding.yaml b/packages/system/hami/charts/hami/templates/device-plugin/monitorrolebinding.yaml new file mode 100644 index 00000000..2f0a14ba --- /dev/null +++ b/packages/system/hami/charts/hami/templates/device-plugin/monitorrolebinding.yaml @@ -0,0 +1,17 @@ +{{- if .Values.devicePlugin.enabled -}} +apiVersion: rbac.authorization.k8s.io/v1 +kind: ClusterRoleBinding +metadata: + name: {{ include "hami-vgpu.device-plugin" . }} + labels: + app.kubernetes.io/component: "hami-device-plugin" + {{- include "hami-vgpu.labels" . | nindent 4 }} +roleRef: + apiGroup: rbac.authorization.k8s.io + kind: ClusterRole + name: {{ include "hami-vgpu.device-plugin" . }}-monitor +subjects: + - kind: ServiceAccount + name: {{ include "hami-vgpu.device-plugin" . }} + namespace: {{ include "hami-vgpu.namespace" . }} +{{- end -}} diff --git a/packages/system/hami/charts/hami/templates/device-plugin/monitorservice.yaml b/packages/system/hami/charts/hami/templates/device-plugin/monitorservice.yaml new file mode 100644 index 00000000..88f1b214 --- /dev/null +++ b/packages/system/hami/charts/hami/templates/device-plugin/monitorservice.yaml @@ -0,0 +1,29 @@ +{{- if .Values.devicePlugin.enabled -}} +apiVersion: v1 +kind: Service +metadata: + name: {{ include "hami-vgpu.device-plugin" . }}-monitor + namespace: {{ include "hami-vgpu.namespace" . }} + labels: + app.kubernetes.io/component: hami-device-plugin + {{- include "hami-vgpu.labels" . | nindent 4 }} + {{- with .Values.devicePlugin.service.labels }} + {{- toYaml . | nindent 4 }} + {{- end }} + {{- if .Values.devicePlugin.service.annotations }} # Use devicePlugin instead of scheduler + annotations: {{ toYaml .Values.devicePlugin.service.annotations | nindent 4 }} + {{- end }} +spec: + type: {{ .Values.devicePlugin.service.type | default "NodePort" }} # Default type is NodePort + ports: + - name: monitorport + port: {{ .Values.devicePlugin.service.httpPort | default 31992 }} # Default HTTP port is 31992 + targetPort: 9394 + {{- if eq (.Values.devicePlugin.service.type | default "NodePort") "NodePort" }} # If type is NodePort, set nodePort + nodePort: {{ .Values.devicePlugin.service.httpPort | default 31992 }} + {{- end }} + protocol: TCP + selector: + app.kubernetes.io/component: hami-device-plugin + {{- include "hami-vgpu.selectorLabels" . | nindent 4 }} +{{- end -}} diff --git a/packages/system/hami/charts/hami/templates/device-plugin/monitorserviceaccount.yaml b/packages/system/hami/charts/hami/templates/device-plugin/monitorserviceaccount.yaml new file mode 100644 index 00000000..6e3c2a44 --- /dev/null +++ b/packages/system/hami/charts/hami/templates/device-plugin/monitorserviceaccount.yaml @@ -0,0 +1,10 @@ +{{- if .Values.devicePlugin.enabled -}} +apiVersion: v1 +kind: ServiceAccount +metadata: + name: {{ include "hami-vgpu.device-plugin" . }} + namespace: {{ include "hami-vgpu.namespace" . }} + labels: + app.kubernetes.io/component: "hami-device-plugin" + {{- include "hami-vgpu.labels" . | nindent 4 }} +{{- end -}} diff --git a/packages/system/hami/charts/hami/templates/device-plugin/runtime-class.yaml b/packages/system/hami/charts/hami/templates/device-plugin/runtime-class.yaml new file mode 100644 index 00000000..ebcc2c9c --- /dev/null +++ b/packages/system/hami/charts/hami/templates/device-plugin/runtime-class.yaml @@ -0,0 +1,9 @@ +{{- if and .Values.devicePlugin.enabled .Values.devicePlugin.createRuntimeClass .Values.devicePlugin.runtimeClassName -}} +apiVersion: node.k8s.io/v1 +kind: RuntimeClass +metadata: + name: {{ .Values.devicePlugin.runtimeClassName }} + annotations: + helm.sh/hook: pre-install,pre-upgrade +handler: nvidia +{{- end -}} diff --git a/packages/system/hami/charts/hami/templates/scheduler/certmanager.yaml b/packages/system/hami/charts/hami/templates/scheduler/certmanager.yaml new file mode 100644 index 00000000..e6d28721 --- /dev/null +++ b/packages/system/hami/charts/hami/templates/scheduler/certmanager.yaml @@ -0,0 +1,31 @@ +{{- if .Values.scheduler.admissionWebhook.enabled -}} +{{- if .Values.scheduler.certManager.enabled }} +apiVersion: cert-manager.io/v1 +kind: Certificate +metadata: + name: {{ include "hami-vgpu.scheduler" . }}-serving-cert + namespace: {{ include "hami-vgpu.namespace" . }} + labels: + app.kubernetes.io/component: hami-scheduler + {{- include "hami-vgpu.labels" . | nindent 4 }} +spec: + dnsNames: + - {{ include "hami-vgpu.scheduler" . }}.{{ include "hami-vgpu.namespace" . }}.svc + - {{ include "hami-vgpu.scheduler" . }}.{{ include "hami-vgpu.namespace" . }}.svc.cluster.local + issuerRef: + kind: Issuer + name: {{ include "hami-vgpu.scheduler" . }}-selfsigned-issuer + secretName: {{ include "hami-vgpu.scheduler.tls" . }} +--- +apiVersion: cert-manager.io/v1 +kind: Issuer +metadata: + name: {{ include "hami-vgpu.scheduler" . }}-selfsigned-issuer + namespace: {{ include "hami-vgpu.namespace" . }} + labels: + app.kubernetes.io/component: hami-scheduler + {{- include "hami-vgpu.labels" . | nindent 4 }} +spec: + selfSigned: {} +{{- end }} +{{- end }} diff --git a/packages/system/hami/charts/hami/templates/scheduler/clusterrole.yaml b/packages/system/hami/charts/hami/templates/scheduler/clusterrole.yaml new file mode 100644 index 00000000..81c4fddf --- /dev/null +++ b/packages/system/hami/charts/hami/templates/scheduler/clusterrole.yaml @@ -0,0 +1,34 @@ +apiVersion: rbac.authorization.k8s.io/v1 +kind: ClusterRole +metadata: + name: {{ include "hami-vgpu.scheduler" . }} + labels: + app.kubernetes.io/component: "hami-scheduler" + {{- include "hami-vgpu.labels" . | nindent 4 }} +rules: + - apiGroups: [""] + resources: ["pods", "configmaps"] + verbs: ["get", "list", "watch", "patch"] + - apiGroups: [""] + resources: ["pods/binding"] + verbs: ["create"] + - apiGroups: [""] + resources: ["nodes"] + verbs: ["get", "list", "patch", "watch"] + - apiGroups: [""] + resources: ["events"] + verbs: ["create", "get", "list"] + - apiGroups: [""] + resources: ["resourcequotas"] + verbs: ["get", "list", "watch"] +--- +{{- if .Values.mockDevicePlugin.enabled }} +apiVersion: rbac.authorization.k8s.io/v1 +kind: ClusterRole +metadata: + name: {{ include "hami-vgpu.mock-device-plugin" . }} +rules: + - apiGroups: [""] + resources: ["nodes"] + verbs: ["get", "update", "list", "patch"] +{{- end -}} diff --git a/packages/system/hami/charts/hami/templates/scheduler/clusterrolebinding.yaml b/packages/system/hami/charts/hami/templates/scheduler/clusterrolebinding.yaml new file mode 100644 index 00000000..a81d425c --- /dev/null +++ b/packages/system/hami/charts/hami/templates/scheduler/clusterrolebinding.yaml @@ -0,0 +1,47 @@ +apiVersion: rbac.authorization.k8s.io/v1 +kind: ClusterRoleBinding +metadata: + name: {{ include "hami-vgpu.scheduler" . }}-kube + labels: + app.kubernetes.io/component: "hami-scheduler" + {{- include "hami-vgpu.labels" . | nindent 4 }} +roleRef: + apiGroup: rbac.authorization.k8s.io + kind: ClusterRole + name: system:kube-scheduler +subjects: + - kind: ServiceAccount + name: {{ include "hami-vgpu.scheduler" . }} + namespace: {{ include "hami-vgpu.namespace" . }} +--- +apiVersion: rbac.authorization.k8s.io/v1 +kind: ClusterRoleBinding +metadata: + name: {{ include "hami-vgpu.scheduler" . }}-volume + labels: + app.kubernetes.io/component: "hami-scheduler" + {{- include "hami-vgpu.labels" . | nindent 4 }} +roleRef: + apiGroup: rbac.authorization.k8s.io + kind: ClusterRole + name: system:volume-scheduler +subjects: + - kind: ServiceAccount + name: {{ include "hami-vgpu.scheduler" . }} + namespace: {{ include "hami-vgpu.namespace" . }} +--- +apiVersion: rbac.authorization.k8s.io/v1 +kind: ClusterRoleBinding +metadata: + name: {{ include "hami-vgpu.scheduler" . }} + labels: + app.kubernetes.io/component: "hami-scheduler" + {{- include "hami-vgpu.labels" . | nindent 4 }} +roleRef: + apiGroup: rbac.authorization.k8s.io + kind: ClusterRole + name: {{ include "hami-vgpu.scheduler" . }} +subjects: + - kind: ServiceAccount + name: {{ include "hami-vgpu.scheduler" . }} + namespace: {{ include "hami-vgpu.namespace" . }} diff --git a/packages/system/hami/charts/hami/templates/scheduler/configmap.yaml b/packages/system/hami/charts/hami/templates/scheduler/configmap.yaml new file mode 100644 index 00000000..109ecbce --- /dev/null +++ b/packages/system/hami/charts/hami/templates/scheduler/configmap.yaml @@ -0,0 +1,142 @@ +{{- if .Values.scheduler.kubeScheduler.enabled -}} +apiVersion: v1 +kind: ConfigMap +metadata: + name: {{ include "hami-vgpu.scheduler" . }} + namespace: {{ include "hami-vgpu.namespace" . }} + labels: + app.kubernetes.io/component: hami-scheduler + {{- include "hami-vgpu.labels" . | nindent 4 }} +data: + config.json: | + { + "kind": "Policy", + "apiVersion": "v1", + "extenders": [ + { + {{- if .Values.scheduler.admissionWebhook.enabled }} + "urlPrefix": "https://127.0.0.1:443", + "enableHttps": true, + "tlsConfig": { + "insecure": true + }, + {{- else }} + "urlPrefix": "http://127.0.0.1:80", + "enableHttps": false, + {{- end }} + "filterVerb": "filter", + "bindVerb": "bind", + "weight": 1, + "nodeCacheCapable": true, + "httpTimeout": 30000000000, + "managedResources": [ + {{- range .Values.devices.amd.customresources }} + { + "name": "{{ . }}", + "ignoredByScheduler": true + }, + {{- end }} + {{- if .Values.devices.ascend.enabled }} + {{- range .Values.devices.ascend.customresources }} + { + "name": "{{ . }}", + "ignoredByScheduler": true + }, + {{- end }} + {{- end }} + {{- if .Values.devices.mthreads.enabled }} + {{- range .Values.devices.mthreads.customresources }} + { + "name": "{{ . }}", + "ignoredByScheduler": true + }, + {{- end }} + {{- end }} + {{- if .Values.devices.enflame.enabled }} + {{- range .Values.devices.enflame.customresources }} + { + "name": "{{ . }}", + "ignoredByScheduler": true + }, + {{- end }} + {{- end }} + {{- if .Values.devices.kunlun.enabled }} + {{- range .Values.devices.kunlun.customresources }} + { + "name": "{{ . }}", + "ignoredByScheduler": true + }, + {{- end }} + {{- end }} + {{- range .Values.devices.awsneuron.customresources }} + { + "name": "{{ . }}", + "ignoredByScheduler": true + }, + {{- end }} + {{- if .Values.devices.iluvatar.enabled }} + {{- range .Values.devices.iluvatar.customresources }} + { + "name": "{{ . }}", + "ignoredByScheduler": true + }, + {{- end }} + {{- end }} + { + "name": "{{ .Values.resourceName }}", + "ignoredByScheduler": true + }, + { + "name": "{{ .Values.resourceMem }}", + "ignoredByScheduler": true + }, + { + "name": "{{ .Values.resourceCores }}", + "ignoredByScheduler": true + }, + { + "name": "{{ .Values.resourceMemPercentage }}", + "ignoredByScheduler": true + }, + { + "name": "{{ .Values.resourcePriority }}", + "ignoredByScheduler": true + }, + { + "name": "{{ .Values.mluResourceName }}", + "ignoredByScheduler": true + }, + { + "name": "{{ .Values.dcuResourceName }}", + "ignoredByScheduler": true + }, + { + "name": "{{ .Values.dcuResourceMem }}", + "ignoredByScheduler": true + }, + { + "name": "{{ .Values.dcuResourceCores }}", + "ignoredByScheduler": true + }, + { + "name": "metax-tech.com/gpu", + "ignoredByScheduler": true + }, + { + "name": "{{ .Values.metaxResourceName }}", + "ignoredByScheduler": true + }, + { + "name": "{{ .Values.metaxResourceCore }}", + "ignoredByScheduler": true + }, + { + "name": "{{ .Values.metaxResourceMem }}", + "ignoredByScheduler": true + } + ], + "ignoreable": false + } + ] + } +{{- end }} diff --git a/packages/system/hami/charts/hami/templates/scheduler/configmapnew.yaml b/packages/system/hami/charts/hami/templates/scheduler/configmapnew.yaml new file mode 100644 index 00000000..6f6db097 --- /dev/null +++ b/packages/system/hami/charts/hami/templates/scheduler/configmapnew.yaml @@ -0,0 +1,102 @@ +{{- if .Values.scheduler.kubeScheduler.enabled -}} +apiVersion: v1 +kind: ConfigMap +metadata: + name: {{ include "hami-vgpu.scheduler" . }}-newversion + namespace: {{ include "hami-vgpu.namespace" . }} + labels: + app.kubernetes.io/component: hami-scheduler + {{- include "hami-vgpu.labels" . | nindent 4 }} +data: + config.yaml: | + {{- if gt (regexReplaceAll "[^0-9]" .Capabilities.KubeVersion.Minor "" | int) 25}} + apiVersion: kubescheduler.config.k8s.io/v1 + {{- else }} + apiVersion: kubescheduler.config.k8s.io/v1beta2 + {{- end }} + kind: KubeSchedulerConfiguration + leaderElection: + leaderElect: false + profiles: + - schedulerName: {{ .Values.schedulerName }} + extenders: + {{- if .Values.scheduler.admissionWebhook.enabled }} + - urlPrefix: "https://127.0.0.1:443" + enableHTTPS: true + tlsConfig: + insecure: true + {{- else }} + - urlPrefix: "http://127.0.0.1:80" + enableHTTPS: false + {{- end }} + filterVerb: filter + bindVerb: bind + nodeCacheCapable: true + weight: 1 + httpTimeout: 30s + managedResources: + - name: {{ .Values.resourceName }} + ignoredByScheduler: true + - name: {{ .Values.resourceMem }} + ignoredByScheduler: true + - name: {{ .Values.resourceCores }} + ignoredByScheduler: true + - name: {{ .Values.resourceMemPercentage }} + ignoredByScheduler: true + - name: {{ .Values.resourcePriority }} + ignoredByScheduler: true + - name: {{ .Values.mluResourceName }} + ignoredByScheduler: true + - name: {{ .Values.dcuResourceName }} + ignoredByScheduler: true + - name: {{ .Values.dcuResourceMem }} + ignoredByScheduler: true + - name: {{ .Values.dcuResourceCores }} + ignoredByScheduler: true + - name: "metax-tech.com/gpu" + ignoredByScheduler: true + - name: {{ .Values.metaxResourceName }} + ignoredByScheduler: true + - name: {{ .Values.metaxResourceCore }} + ignoredByScheduler: true + - name: {{ .Values.metaxResourceMem }} + ignoredByScheduler: true + {{- if .Values.devices.ascend.enabled }} + {{- range .Values.devices.ascend.customresources }} + - name: {{ . }} + ignoredByScheduler: true + {{- end }} + {{- end }} + {{- if .Values.devices.mthreads.enabled }} + {{- range .Values.devices.mthreads.customresources }} + - name: {{ . }} + ignoredByScheduler: true + {{- end }} + {{- end }} + {{- if .Values.devices.enflame.enabled }} + {{- range .Values.devices.enflame.customresources }} + - name: {{ . }} + ignoredByScheduler: true + {{- end }} + {{- end }} + {{- if .Values.devices.kunlun.enabled }} + {{- range .Values.devices.kunlun.customresources }} + - name: {{ . }} + ignoredByScheduler: true + {{- end }} + {{- end }} + {{- range .Values.devices.awsneuron.customresources }} + - name: {{ . }} + ignoredByScheduler: true + {{- end }} + {{- if .Values.devices.iluvatar.enabled }} + {{- range .Values.devices.iluvatar.customresources }} + - name: {{ . }} + ignoredByScheduler: true + {{- end }} + {{- end }} + {{- range .Values.devices.amd.customresources }} + - name: {{ . }} + ignoredByScheduler: true + {{- end }} +{{- end }} diff --git a/packages/system/hami/charts/hami/templates/scheduler/deployment.yaml b/packages/system/hami/charts/hami/templates/scheduler/deployment.yaml new file mode 100644 index 00000000..1d89e189 --- /dev/null +++ b/packages/system/hami/charts/hami/templates/scheduler/deployment.yaml @@ -0,0 +1,225 @@ +apiVersion: apps/v1 +kind: Deployment +metadata: + name: {{ include "hami-vgpu.scheduler" . }} + namespace: {{ include "hami-vgpu.namespace" . }} + labels: + app.kubernetes.io/component: hami-scheduler + {{- include "hami-vgpu.labels" . | nindent 4 }} + {{- with .Values.global.labels }} + {{- toYaml . | nindent 4 }} + {{- end }} + {{- if .Values.global.annotations }} + annotations: {{ toYaml .Values.global.annotations | nindent 4}} + {{- end }} +spec: + {{- if .Values.scheduler.leaderElect }} + replicas: {{ .Values.scheduler.replicas }} + {{- else }} + replicas: 1 + {{- end }} + selector: + matchLabels: + app.kubernetes.io/component: hami-scheduler + {{- include "hami-vgpu.selectorLabels" . | nindent 6 }} + template: + metadata: + labels: + app.kubernetes.io/component: hami-scheduler + {{- include "hami-vgpu.selectorLabels" . | nindent 8 }} + hami.io/webhook: ignore + annotations: + {{- if ge (regexReplaceAll "[^0-9]" .Capabilities.KubeVersion.Minor "" | int) 22 }} + checksum/hami-scheduler-newversion-config: {{ include (print $.Template.BasePath "/scheduler/configmapnew.yaml") . | sha256sum }} + {{- else }} + checksum/hami-scheduler-config: {{ include (print $.Template.BasePath "/scheduler/configmap.yaml") . | sha256sum }} + {{- end }} + checksum/hami-scheduler-device-config: {{ include (print $.Template.BasePath "/scheduler/device-configmap.yaml") . | sha256sum }} + {{- if .Values.scheduler.podAnnotations }} + {{- toYaml .Values.scheduler.podAnnotations | nindent 8 }} + {{- end }} + spec: + serviceAccountName: {{ include "hami-vgpu.scheduler" . }} + priorityClassName: system-node-critical + {{- include "hami.scheduler.extender.imagePullSecrets" . | nindent 6 }} + containers: + {{- if .Values.scheduler.kubeScheduler.enabled }} + - name: kube-scheduler + image: {{ include "hami.scheduler.kubeScheduler.image" . }} + imagePullPolicy: {{ .Values.scheduler.kubeScheduler.image.pullPolicy }} + command: + - kube-scheduler + {{- if ge (regexReplaceAll "[^0-9]" .Capabilities.KubeVersion.Minor "" | int) 22 }} + {{- range .Values.scheduler.kubeScheduler.extraNewArgs }} + - {{ . }} + {{- end }} + {{- else }} + - --scheduler-name={{ .Values.schedulerName }} + {{- range .Values.scheduler.kubeScheduler.extraArgs }} + - {{ . }} + {{- end }} + {{- end }} + - --leader-elect={{ .Values.scheduler.leaderElect }} + - --leader-elect-resource-name={{ .Values.schedulerName }} + - --leader-elect-resource-namespace={{ include "hami-vgpu.namespace" . }} + resources: + {{- toYaml .Values.scheduler.kubeScheduler.resources | nindent 12 }} + volumeMounts: + - name: scheduler-config + mountPath: /config + {{- end }} + {{- if .Values.scheduler.livenessProbe }} + livenessProbe: + failureThreshold: 8 + httpGet: + path: /healthz + port: 10259 + scheme: HTTPS + initialDelaySeconds: 10 + periodSeconds: 10 + successThreshold: 1 + timeoutSeconds: 15 + {{- end }} + - name: vgpu-scheduler-extender + image: {{ include "hami.scheduler.extender.image" . }} + imagePullPolicy: {{ .Values.scheduler.extender.image.pullPolicy }} + env: + {{- if .Values.scheduler.nodeLockExpire }} + - name: HAMI_NODELOCK_EXPIRE + value: "{{ .Values.scheduler.nodeLockExpire }}" + {{- end }} + {{- if .Values.global.managedNodeSelectorEnable }} + {{- range $key, $value := .Values.global.managedNodeSelector }} + - name: NODE_SELECTOR_{{ $key | upper | replace "-" "_" }} + value: "{{ $value }}" + {{- end }} + {{- end }} + command: + - scheduler + {{- if .Values.scheduler.admissionWebhook.enabled }} + - --http_bind=0.0.0.0:443 + - --cert_file=/tls/tls.crt + - --key_file=/tls/tls.key + {{- else }} + - --http_bind=0.0.0.0:80 + {{- end }} + - --scheduler-name={{ .Values.schedulerName }} + - --metrics-bind-address={{ .Values.scheduler.metricsBindAddress }} + - --node-scheduler-policy={{ .Values.scheduler.defaultSchedulerPolicy.nodeSchedulerPolicy }} + - --gpu-scheduler-policy={{ .Values.scheduler.defaultSchedulerPolicy.gpuSchedulerPolicy }} + - --force-overwrite-default-scheduler={{ .Values.scheduler.forceOverwriteDefaultScheduler}} + - --device-config-file=/device-config.yaml + - --leader-elect={{ .Values.scheduler.leaderElect }} + - --leader-elect-resource-name={{ .Values.schedulerName }} + - --leader-elect-resource-namespace={{ include "hami-vgpu.namespace" . }} + {{- if .Values.devices.ascend.enabled }} + - --enable-ascend=true + {{- end }} + {{- if .Values.devices.iluvatar.enabled }} + - --enable-iluvatar=true + {{- end }} + {{- if .Values.scheduler.nodeLabelSelector }} + - --node-label-selector={{- $first := true -}} + {{- range $key, $value := .Values.scheduler.nodeLabelSelector -}} + {{- if not $first }},{{ end -}} + {{- $key }}={{ $value -}} + {{- $first = false -}} + {{- end -}} + {{- end }} + {{- range .Values.scheduler.extender.extraArgs }} + - {{ . }} + {{- end }} + ports: + {{- if .Values.scheduler.admissionWebhook.enabled }} + - name: https + containerPort: 443 + protocol: TCP + {{- else }} + - name: http + containerPort: 80 + protocol: TCP + {{- end }} + - name: metrics + containerPort: {{ last (splitList ":" .Values.scheduler.metricsBindAddress) | int }} + protocol: TCP + resources: + {{- toYaml .Values.scheduler.extender.resources | nindent 12 }} + volumeMounts: + - name: device-config + mountPath: /device-config.yaml + subPath: device-config.yaml + {{- if .Values.scheduler.admissionWebhook.enabled }} + - name: tls-config + mountPath: /tls + {{- end }} + {{- if .Values.scheduler.livenessProbe }} + livenessProbe: + httpGet: + path: /healthz + {{- if .Values.scheduler.admissionWebhook.enabled }} + port: https + scheme: HTTPS + {{- else }} + port: http + scheme: HTTP + {{- end }} + initialDelaySeconds: 10 + periodSeconds: 10 + failureThreshold: 3 + timeoutSeconds: 5 + {{- end }} + {{- if .Values.scheduler.leaderElect }} + readinessProbe: + httpGet: + path: /readyz + {{- if .Values.scheduler.admissionWebhook.enabled }} + port: https + scheme: HTTPS + {{- else }} + port: http + scheme: HTTP + {{- end }} + initialDelaySeconds: 10 + periodSeconds: 10 + failureThreshold: 3 + timeoutSeconds: 5 + {{- end }} + {{- if .Values.scheduler.leaderElect }} + affinity: + podAntiAffinity: + requiredDuringSchedulingIgnoredDuringExecution: + - labelSelector: + matchExpressions: + - key: app.kubernetes.io/component + operator: In + values: + - hami-scheduler + topologyKey: "kubernetes.io/hostname" + {{- end }} + volumes: + {{- if .Values.scheduler.admissionWebhook.enabled }} + - name: tls-config + secret: + secretName: {{ template "hami-vgpu.scheduler.tls" . }} + {{- end }} + {{- if .Values.scheduler.kubeScheduler.enabled }} + - name: scheduler-config + configMap: + {{- if ge (regexReplaceAll "[^0-9]" .Capabilities.KubeVersion.Minor "" | int) 22 }} + name: {{ template "hami-vgpu.scheduler" . }}-newversion + {{- else }} + name: {{ template "hami-vgpu.scheduler" . }} + {{- end }} + {{- end }} + - name: device-config + configMap: + name: {{ include "hami-vgpu.scheduler" . }}-device + {{- if .Values.scheduler.nodeSelector }} + nodeSelector: {{ toYaml .Values.scheduler.nodeSelector | nindent 8 }} + {{- end }} + {{- if .Values.scheduler.tolerations }} + tolerations: {{ toYaml .Values.scheduler.tolerations | nindent 8 }} + {{- end }} + {{- if .Values.scheduler.nodeName }} + nodeName: {{ .Values.scheduler.nodeName }} + {{- end }} diff --git a/packages/system/hami/charts/hami/templates/scheduler/device-configmap.yaml b/packages/system/hami/charts/hami/templates/scheduler/device-configmap.yaml new file mode 100644 index 00000000..873b813b --- /dev/null +++ b/packages/system/hami/charts/hami/templates/scheduler/device-configmap.yaml @@ -0,0 +1,408 @@ +apiVersion: v1 +kind: ConfigMap +metadata: + name: {{ include "hami-vgpu.scheduler" . }}-device + namespace: {{ include "hami-vgpu.namespace" . }} + labels: + app.kubernetes.io/component: hami-scheduler + {{- include "hami-vgpu.labels" . | nindent 4 }} +data: + device-config.yaml: |- + {{- if .Files.Glob "files/device-config.yaml" }} + {{- .Files.Get "files/device-config.yaml" | nindent 4}} + {{- else }} + nvidia: + resourceCountName: {{ .Values.resourceName }} + resourceMemoryName: {{ .Values.resourceMem }} + resourceMemoryPercentageName: {{ .Values.resourceMemPercentage }} + resourceCoreName: {{ .Values.resourceCores }} + resourcePriorityName: {{ .Values.resourcePriority }} + overwriteEnv: false + defaultMemory: 0 + defaultCores: 0 + defaultGPUNum: 1 + memoryFactor: 1 + deviceSplitCount: {{ .Values.devicePlugin.deviceSplitCount }} + deviceMemoryScaling: {{ .Values.devicePlugin.deviceMemoryScaling }} + deviceCoreScaling: {{ .Values.devicePlugin.deviceCoreScaling }} + gpuCorePolicy: {{ .Values.devices.nvidia.gpuCorePolicy }} + libCudaLogLevel: {{ .Values.devices.nvidia.libCudaLogLevel }} + runtimeClassName: "{{ .Values.devicePlugin.runtimeClassName }}" + knownMigGeometries: + - models: [ "A30" ] + allowedGeometries: + - + - name: 1g.6gb + core: 25 + memory: 6144 + count: 4 + - + - name: 2g.12gb + core: 50 + memory: 12288 + count: 2 + - + - name: 4g.24gb + core: 100 + memory: 24576 + count: 1 + - models: [ "A100-SXM4-40GB", "A100-40GB-PCIe", "A100-PCIE-40GB"] + allowedGeometries: + - + - name: 1g.5gb + core: 14 + memory: 5120 + count: 7 + - + - name: 1g.5gb + core: 14 + memory: 5120 + count: 1 + - name: 2g.10gb + core: 28 + memory: 10240 + count: 3 + - + - name: 3g.20gb + core: 42 + memory: 20480 + count: 2 + - + - name: 7g.40gb + core: 100 + memory: 40960 + count: 1 + - models: [ "A100-SXM4-80GB", "A100-80GB-PCIe", "A100-PCIE-80GB"] + allowedGeometries: + - + - name: 1g.10gb + core: 14 + memory: 10240 + count: 7 + - + - name: 1g.10gb + core: 14 + memory: 10240 + count: 1 + - name: 2g.20gb + core: 28 + memory: 20480 + count: 3 + - + - name: 3g.40gb + core: 42 + memory: 40960 + count: 2 + - + - name: 7g.79gb + core: 100 + memory: 80896 + count: 1 + - models: [ "H100-PCIE-80GB", "H100-SXM5-80GB"] + allowedGeometries: + - + - name: 1g.10gb + core: 14 + memory: 10240 + count: 7 + - + - name: 1g.10gb + core: 14 + memory: 10240 + count: 1 + - name: 2g.20gb + core: 28 + memory: 20480 + count: 3 + - + - name: 3g.40gb + core: 42 + memory: 40960 + count: 2 + - + - name: 7g.80gb + core: 100 + memory: 81920 + count: 1 + - models: [ "H100-PCIE-94GB", "H100-SXM5-94GB"] + allowedGeometries: + - + - name: 1g.12gb + core: 14 + memory: 12288 + count: 7 + - + - name: 1g.12gb + core: 14 + memory: 12288 + count: 1 + - name: 2g.24gb + core: 28 + memory: 24576 + count: 3 + - + - name: 3g.47gb + core: 42 + memory: 48128 + count: 2 + - + - name: 7g.94gb + core: 100 + memory: 96256 + count: 1 + - models: [ "H20", "H100 on GH200"] + allowedGeometries: + - + - name: 1g.12gb + core: 14 + memory: 12288 + count: 7 + - + - name: 1g.12gb + core: 14 + memory: 12288 + count: 1 + - name: 2g.24gb + core: 28 + memory: 24576 + count: 3 + - + - name: 3g.48gb + core: 42 + memory: 49152 + count: 2 + - + - name: 7g.96gb + core: 100 + memory: 98304 + count: 1 + - models: [ "H200 NVL", "H200-SXM5"] + allowedGeometries: + - + - name: 1g.18gb + core: 14 + memory: 18432 + count: 7 + - + - name: 1g.18gb + core: 14 + memory: 18432 + count: 1 + - name: 2g.35gb + core: 28 + memory: 35840 + count: 3 + - + - name: 3g.71gb + core: 42 + memory: 72704 + count: 2 + - + - name: 7g.141gb + core: 100 + memory: 144384 + count: 1 + - models: [ "B200" ] + allowedGeometries: + - + - name: 1g.23gb + core: 14 + memory: 23552 + count: 7 + - + - name: 1g.23gb + core: 14 + memory: 23552 + count: 1 + - name: 2g.45gb + core: 28 + memory: 46080 + count: 3 + - + - name: 3g.90gb + core: 42 + memory: 92160 + count: 2 + - + - name: 7g.180gb + core: 100 + memory: 184320 + count: 1 + cambricon: + resourceCountName: {{ .Values.mluResourceName }} + resourceMemoryName: {{ .Values.mluResourceMem }} + resourceCoreName: {{ .Values.mluResourceCores }} + hygon: + resourceCountName: {{ .Values.dcuResourceName }} + resourceMemoryName: {{ .Values.dcuResourceMem }} + resourceCoreName: {{ .Values.dcuResourceCores }} + memoryFactor: 1 + metax: + resourceCountName: "metax-tech.com/gpu" + resourceVCountName: {{ .Values.metaxResourceName }} + resourceVMemoryName: {{ .Values.metaxResourceMem }} + resourceVCoreName: {{ .Values.metaxResourceCore }} + sgpuTopologyAware: {{ .Values.metaxsGPUTopologyAware }} + enflame: + resourceNameGCU: "enflame.com/gcu" + resourceNameVGCU: {{ .Values.enflameResourceNameVGCU }} + resourceNameVGCUPercentage: {{ .Values.enflameResourceNameVGCUPercentage }} + mthreads: + resourceCountName: "mthreads.com/vgpu" + resourceMemoryName: "mthreads.com/sgpu-memory" + resourceCoreName: "mthreads.com/sgpu-core" + iluvatars: + - chipName: MR-V100 + commonWord: MR-V100 + resourceCountName: iluvatar.ai/MR-V100-vgpu + resourceMemoryName: iluvatar.ai/MR-V100.vMem + resourceCoreName: iluvatar.ai/MR-V100.vCore + - chipName: MR-V50 + commonWord: MR-V50 + resourceCountName: iluvatar.ai/MR-V50-vgpu + resourceMemoryName: iluvatar.ai/MR-V50.vMem + resourceCoreName: iluvatar.ai/MR-V50.vCore + - chipName: BI-V150 + commonWord: BI-V150 + resourceCountName: iluvatar.ai/BI-V150-vgpu + resourceMemoryName: iluvatar.ai/BI-V150.vMem + resourceCoreName: iluvatar.ai/BI-V150.vCore + - chipName: BI-V100 + commonWord: BI-V100 + resourceCountName: iluvatar.ai/BI-V100-vgpu + resourceMemoryName: iluvatar.ai/BI-V100.vMem + resourceCoreName: iluvatar.ai/BI-V100.vCore + kunlun: + resourceCountName: {{ .Values.kunlunResourceName }} + resourceVCountName: {{ .Values.kunlunResourceVCountName }} + resourceVMemoryName: {{ .Values.kunlunResourceVMemoryName }} + awsneuron: + resourceCountName: "aws.amazon.com/neuron" + resourceCoreName: "aws.amazon.com/neuroncore" + amd: + resourceCountName: "amd.com/gpu" + vnpus: + - chipName: 910A + commonWord: Ascend910A + resourceName: huawei.com/Ascend910A + resourceMemoryName: huawei.com/Ascend910A-memory + memoryAllocatable: 32768 + memoryCapacity: 32768 + memoryFactor: 1 + aiCore: 30 + templates: + - name: vir02 + memory: 2184 + aiCore: 2 + - name: vir04 + memory: 4369 + aiCore: 4 + - name: vir08 + memory: 8738 + aiCore: 8 + - name: vir16 + memory: 17476 + aiCore: 16 + - chipName: 910B2 + commonWord: Ascend910B2 + resourceName: huawei.com/Ascend910B2 + resourceMemoryName: huawei.com/Ascend910B2-memory + memoryAllocatable: 65536 + memoryCapacity: 65536 + memoryFactor: 1 + aiCore: 24 + aiCPU: 6 + templates: + - name: vir03_1c_8g + memory: 8192 + aiCore: 3 + aiCPU: 1 + - name: vir06_1c_16g + memory: 16384 + aiCore: 6 + aiCPU: 1 + - name: vir12_3c_32g + memory: 32768 + aiCore: 12 + aiCPU: 3 + - chipName: 910B3 + commonWord: Ascend910B3 + resourceName: huawei.com/Ascend910B3 + resourceMemoryName: huawei.com/Ascend910B3-memory + memoryAllocatable: 65536 + memoryCapacity: 65536 + memoryFactor: 1 + aiCore: 20 + aiCPU: 7 + templates: + - name: vir05_1c_16g + memory: 16384 + aiCore: 5 + aiCPU: 1 + - name: vir10_3c_32g + memory: 32768 + aiCore: 10 + aiCPU: 3 + - chipName: 910B4-1 + commonWord: Ascend910B4-1 + resourceName: huawei.com/Ascend910B4-1 + resourceMemoryName: huawei.com/Ascend910B4-1-memory + memoryAllocatable: 65536 + memoryCapacity: 65536 + memoryFactor: 1 + aiCore: 20 + aiCPU: 7 + templates: + # NOTE: Names of vnpu template for 910B4-1 are fixed by Ascend runtime and must not be changed. + # The memory is used for scheduling so the correct values must be set. + # Template vir05_1c_8g actually provides 16GB memory, + - name: vir05_1c_8g + memory: 16384 + aiCore: 5 + aiCPU: 1 + # Template vir10_3c_16g actually provides 32GB memory + - name: vir10_3c_16g + memory: 32768 + aiCore: 10 + aiCPU: 3 + - chipName: 910B4 + commonWord: Ascend910B4 + resourceName: huawei.com/Ascend910B4 + resourceMemoryName: huawei.com/Ascend910B4-memory + memoryAllocatable: 32768 + memoryCapacity: 32768 + memoryFactor: 1 + aiCore: 20 + aiCPU: 7 + templates: + - name: vir05_1c_8g + memory: 8192 + aiCore: 5 + aiCPU: 1 + - name: vir10_3c_16g + memory: 16384 + aiCore: 10 + aiCPU: 3 + - chipName: 310P3 + commonWord: Ascend310P + resourceName: huawei.com/Ascend310P + resourceMemoryName: huawei.com/Ascend310P-memory + memoryAllocatable: 21527 + memoryCapacity: 24576 + memoryFactor: 1 + aiCore: 8 + aiCPU: 7 + templates: + - name: vir01 + memory: 3072 + aiCore: 1 + aiCPU: 1 + - name: vir02 + memory: 6144 + aiCore: 2 + aiCPU: 2 + - name: vir04 + memory: 12288 + aiCore: 4 + aiCPU: 4 + {{ end }} diff --git a/packages/system/hami/charts/hami/templates/scheduler/job-patch/clusterrole.yaml b/packages/system/hami/charts/hami/templates/scheduler/job-patch/clusterrole.yaml new file mode 100644 index 00000000..77e891cc --- /dev/null +++ b/packages/system/hami/charts/hami/templates/scheduler/job-patch/clusterrole.yaml @@ -0,0 +1,21 @@ +{{- if and .Values.scheduler.admissionWebhook.enabled (.Values.scheduler.patch.enabled) (not .Values.scheduler.certManager.enabled) -}} +apiVersion: rbac.authorization.k8s.io/v1 +kind: ClusterRole +metadata: + name: {{ include "hami-vgpu.fullname" . }}-admission + annotations: + "helm.sh/hook": pre-install,pre-upgrade,post-install,post-upgrade + "helm.sh/hook-delete-policy": before-hook-creation,hook-succeeded + labels: + {{- include "hami-vgpu.labels" . | nindent 4 }} + app.kubernetes.io/component: admission-webhook +rules: + - apiGroups: + - admissionregistration.k8s.io + resources: + #- validatingwebhookconfigurations + - mutatingwebhookconfigurations + verbs: + - get + - update +{{- end }} diff --git a/packages/system/hami/charts/hami/templates/scheduler/job-patch/clusterrolebinding.yaml b/packages/system/hami/charts/hami/templates/scheduler/job-patch/clusterrolebinding.yaml new file mode 100644 index 00000000..2b82f926 --- /dev/null +++ b/packages/system/hami/charts/hami/templates/scheduler/job-patch/clusterrolebinding.yaml @@ -0,0 +1,20 @@ +{{- if and .Values.scheduler.admissionWebhook.enabled (.Values.scheduler.patch.enabled) (not .Values.scheduler.certManager.enabled) -}} +apiVersion: rbac.authorization.k8s.io/v1 +kind: ClusterRoleBinding +metadata: + name: {{ include "hami-vgpu.fullname" . }}-admission + annotations: + "helm.sh/hook": pre-install,pre-upgrade,post-install,post-upgrade + "helm.sh/hook-delete-policy": before-hook-creation,hook-succeeded + labels: + {{- include "hami-vgpu.labels" . | nindent 4 }} + app.kubernetes.io/component: admission-webhook +roleRef: + apiGroup: rbac.authorization.k8s.io + kind: ClusterRole + name: {{ include "hami-vgpu.fullname" . }}-admission +subjects: + - kind: ServiceAccount + name: {{ include "hami-vgpu.fullname" . }}-admission + namespace: {{ include "hami-vgpu.namespace" . }} +{{- end }} diff --git a/packages/system/hami/charts/hami/templates/scheduler/job-patch/job-createSecret.yaml b/packages/system/hami/charts/hami/templates/scheduler/job-patch/job-createSecret.yaml new file mode 100644 index 00000000..0e36d95f --- /dev/null +++ b/packages/system/hami/charts/hami/templates/scheduler/job-patch/job-createSecret.yaml @@ -0,0 +1,68 @@ +{{- if and .Values.scheduler.admissionWebhook.enabled (.Values.scheduler.patch.enabled) (not .Values.scheduler.certManager.enabled) -}} +apiVersion: batch/v1 +kind: Job +metadata: + name: {{ include "hami-vgpu.fullname" . }}-admission-create + namespace: {{ include "hami-vgpu.namespace" . }} + annotations: + "helm.sh/hook": pre-install,pre-upgrade + "helm.sh/hook-delete-policy": before-hook-creation,hook-succeeded + labels: + {{- include "hami-vgpu.labels" . | nindent 4 }} + app.kubernetes.io/component: admission-webhook +spec: + {{- if .Capabilities.APIVersions.Has "batch/v1alpha1" }} + # Alpha feature since k8s 1.12 + ttlSecondsAfterFinished: 0 + {{- end }} + template: + metadata: + name: {{ include "hami-vgpu.fullname" . }}-admission-create + {{- if .Values.scheduler.patch.podAnnotations }} + annotations: {{ toYaml .Values.scheduler.patch.podAnnotations | nindent 8 }} + {{- end }} + labels: + {{- include "hami-vgpu.labels" . | nindent 8 }} + app.kubernetes.io/component: admission-webhook + hami.io/webhook: ignore + spec: + {{- if .Values.scheduler.patch.priorityClassName }} + priorityClassName: {{ .Values.scheduler.patch.priorityClassName }} + {{- end }} + {{- if ge (regexReplaceAll "[^0-9]" .Capabilities.KubeVersion.Minor "" | int) 22 }} + {{- include "hami.scheduler.patch.new.imagePullSecrets" . | nindent 6 }} + {{- else }} + {{- include "hami.scheduler.patch.imagePullSecrets" . | nindent 6 }} + {{- end }} + containers: + - name: create + {{- if ge (regexReplaceAll "[^0-9]" .Capabilities.KubeVersion.Minor "" | int) 22 }} + image: {{ include "hami.scheduler.patch.new.image" . }} + imagePullPolicy: {{ .Values.scheduler.patch.imageNew.pullPolicy }} + {{- else }} + image: {{ include "hami.scheduler.patch.image" . }} + imagePullPolicy: {{ .Values.scheduler.patch.image.pullPolicy }} + {{- end }} + args: + - create + - --cert-name=tls.crt + - --key-name=tls.key + {{- if .Values.scheduler.admissionWebhook.customURL.enabled }} + - --host={{ printf "%s.%s.svc,127.0.0.1,%s" (include "hami-vgpu.scheduler" .) (include "hami-vgpu.namespace" .) .Values.scheduler.admissionWebhook.customURL.host}} + {{- else }} + - --host={{ printf "%s.%s.svc,127.0.0.1" (include "hami-vgpu.scheduler" .) (include "hami-vgpu.namespace" .) }} + {{- end }} + - --namespace={{ include "hami-vgpu.namespace" . }} + - --secret-name={{ include "hami-vgpu.scheduler.tls" . }} + restartPolicy: OnFailure + serviceAccountName: {{ include "hami-vgpu.fullname" . }}-admission + {{- if .Values.scheduler.patch.nodeSelector }} + nodeSelector: {{ toYaml .Values.scheduler.patch.nodeSelector | nindent 8 }} + {{- end }} + {{- if .Values.scheduler.patch.tolerations }} + tolerations: {{ toYaml .Values.scheduler.patch.tolerations | nindent 8 }} + {{- end }} + securityContext: + runAsNonRoot: true + runAsUser: {{ .Values.scheduler.patch.runAsUser }} +{{- end }} diff --git a/packages/system/hami/charts/hami/templates/scheduler/job-patch/job-patchWebhook.yaml b/packages/system/hami/charts/hami/templates/scheduler/job-patch/job-patchWebhook.yaml new file mode 100644 index 00000000..ce52042c --- /dev/null +++ b/packages/system/hami/charts/hami/templates/scheduler/job-patch/job-patchWebhook.yaml @@ -0,0 +1,63 @@ +{{- if and .Values.scheduler.admissionWebhook.enabled (.Values.scheduler.patch.enabled) (not .Values.scheduler.certManager.enabled) -}} +apiVersion: batch/v1 +kind: Job +metadata: + name: {{ include "hami-vgpu.fullname" . }}-admission-patch + namespace: {{ include "hami-vgpu.namespace" . }} + annotations: + "helm.sh/hook": post-install,post-upgrade + "helm.sh/hook-delete-policy": before-hook-creation,hook-succeeded + labels: + {{- include "hami-vgpu.labels" . | nindent 4 }} + app.kubernetes.io/component: admission-webhook +spec: + {{- if .Capabilities.APIVersions.Has "batch/v1alpha1" }} + # Alpha feature since k8s 1.12 + ttlSecondsAfterFinished: 0 + {{- end }} + template: + metadata: + name: {{ include "hami-vgpu.fullname" . }}-admission-patch + {{- if .Values.scheduler.patch.podAnnotations }} + annotations: {{ toYaml .Values.scheduler.patch.podAnnotations | nindent 8 }} + {{- end }} + labels: + {{- include "hami-vgpu.labels" . | nindent 8 }} + app.kubernetes.io/component: admission-webhook + hami.io/webhook: ignore + spec: + {{- if .Values.scheduler.patch.priorityClassName }} + priorityClassName: {{ .Values.scheduler.patch.priorityClassName }} + {{- end }} + {{- if ge (regexReplaceAll "[^0-9]" .Capabilities.KubeVersion.Minor "" | int) 22 }} + {{- include "hami.scheduler.patch.new.imagePullSecrets" . | nindent 6 }} + {{- else }} + {{- include "hami.scheduler.patch.imagePullSecrets" . | nindent 6 }} + {{- end }} + containers: + - name: patch + {{- if ge (regexReplaceAll "[^0-9]" .Capabilities.KubeVersion.Minor "" | int) 22 }} + image: {{ include "hami.scheduler.patch.new.image" . }} + imagePullPolicy: {{ .Values.scheduler.patch.imageNew.pullPolicy }} + {{- else }} + image: {{ include "hami.scheduler.patch.image" . }} + imagePullPolicy: {{ .Values.scheduler.patch.image.pullPolicy }} + {{- end }} + args: + - patch + - --webhook-name={{ include "hami-vgpu.scheduler.webhook" . }} + - --namespace={{ include "hami-vgpu.namespace" . }} + - --patch-validating=false + - --secret-name={{ include "hami-vgpu.scheduler.tls" . }} + restartPolicy: OnFailure + serviceAccountName: {{ include "hami-vgpu.fullname" . }}-admission + {{- if .Values.scheduler.patch.nodeSelector }} + nodeSelector: {{ toYaml .Values.scheduler.patch.nodeSelector | nindent 8 }} + {{- end }} + {{- if .Values.scheduler.patch.tolerations }} + tolerations: {{ toYaml .Values.scheduler.patch.tolerations | nindent 8 }} + {{- end }} + securityContext: + runAsNonRoot: true + runAsUser: {{ .Values.scheduler.patch.runAsUser }} +{{- end }} diff --git a/packages/system/hami/charts/hami/templates/scheduler/job-patch/role.yaml b/packages/system/hami/charts/hami/templates/scheduler/job-patch/role.yaml new file mode 100644 index 00000000..56682f8e --- /dev/null +++ b/packages/system/hami/charts/hami/templates/scheduler/job-patch/role.yaml @@ -0,0 +1,21 @@ +{{- if and .Values.scheduler.admissionWebhook.enabled (.Values.scheduler.patch.enabled) (not .Values.scheduler.certManager.enabled) -}} +apiVersion: rbac.authorization.k8s.io/v1 +kind: Role +metadata: + name: {{ include "hami-vgpu.fullname" . }}-admission + namespace: {{ include "hami-vgpu.namespace" . }} + annotations: + "helm.sh/hook": pre-install,pre-upgrade,post-install,post-upgrade + "helm.sh/hook-delete-policy": before-hook-creation,hook-succeeded + labels: + {{- include "hami-vgpu.labels" . | nindent 4 }} + app.kubernetes.io/component: admission-webhook +rules: + - apiGroups: + - "" + resources: + - secrets + verbs: + - get + - create +{{- end }} diff --git a/packages/system/hami/charts/hami/templates/scheduler/job-patch/rolebinding.yaml b/packages/system/hami/charts/hami/templates/scheduler/job-patch/rolebinding.yaml new file mode 100644 index 00000000..7239b128 --- /dev/null +++ b/packages/system/hami/charts/hami/templates/scheduler/job-patch/rolebinding.yaml @@ -0,0 +1,21 @@ +{{- if and .Values.scheduler.admissionWebhook.enabled (.Values.scheduler.patch.enabled) (not .Values.scheduler.certManager.enabled) -}} +apiVersion: rbac.authorization.k8s.io/v1 +kind: RoleBinding +metadata: + name: {{ include "hami-vgpu.fullname" . }}-admission + namespace: {{ include "hami-vgpu.namespace" . }} + annotations: + "helm.sh/hook": pre-install,pre-upgrade,post-install,post-upgrade + "helm.sh/hook-delete-policy": before-hook-creation,hook-succeeded + labels: + {{- include "hami-vgpu.labels" . | nindent 4 }} + app.kubernetes.io/component: admission-webhook +roleRef: + apiGroup: rbac.authorization.k8s.io + kind: Role + name: {{ include "hami-vgpu.fullname" . }}-admission +subjects: + - kind: ServiceAccount + name: {{ include "hami-vgpu.fullname" . }}-admission + namespace: {{ include "hami-vgpu.namespace" . }} +{{- end }} diff --git a/packages/system/hami/charts/hami/templates/scheduler/job-patch/serviceaccount.yaml b/packages/system/hami/charts/hami/templates/scheduler/job-patch/serviceaccount.yaml new file mode 100644 index 00000000..857c6a8d --- /dev/null +++ b/packages/system/hami/charts/hami/templates/scheduler/job-patch/serviceaccount.yaml @@ -0,0 +1,13 @@ +{{- if and .Values.scheduler.admissionWebhook.enabled (.Values.scheduler.patch.enabled) (not .Values.scheduler.certManager.enabled) }} +apiVersion: v1 +kind: ServiceAccount +metadata: + name: {{ include "hami-vgpu.fullname" . }}-admission + namespace: {{ include "hami-vgpu.namespace" . }} + annotations: + "helm.sh/hook": pre-install,pre-upgrade,post-install,post-upgrade + "helm.sh/hook-delete-policy": before-hook-creation,hook-succeeded + labels: + {{- include "hami-vgpu.labels" . | nindent 4 }} + app.kubernetes.io/component: admission-webhook +{{- end }} diff --git a/packages/system/hami/charts/hami/templates/scheduler/role.yaml b/packages/system/hami/charts/hami/templates/scheduler/role.yaml new file mode 100644 index 00000000..5f7d7e44 --- /dev/null +++ b/packages/system/hami/charts/hami/templates/scheduler/role.yaml @@ -0,0 +1,12 @@ +apiVersion: rbac.authorization.k8s.io/v1 +kind: Role +metadata: + name: {{ include "hami-vgpu.scheduler" . }} + namespace: {{ include "hami-vgpu.namespace" . }} + labels: + app.kubernetes.io/component: "hami-scheduler" + {{- include "hami-vgpu.labels" . | nindent 4 }} +rules: + - apiGroups: ["coordination.k8s.io"] + resources: ["leases"] + verbs: ["create", "list", "watch", "get", "update", "patch"] diff --git a/packages/system/hami/charts/hami/templates/scheduler/rolebinding.yaml b/packages/system/hami/charts/hami/templates/scheduler/rolebinding.yaml new file mode 100644 index 00000000..96e175a1 --- /dev/null +++ b/packages/system/hami/charts/hami/templates/scheduler/rolebinding.yaml @@ -0,0 +1,31 @@ +apiVersion: rbac.authorization.k8s.io/v1 +kind: RoleBinding +metadata: + name: {{ include "hami-vgpu.scheduler" . }} + namespace: {{ include "hami-vgpu.namespace" . }} + labels: + app.kubernetes.io/component: "hami-scheduler" + {{- include "hami-vgpu.labels" . | nindent 4 }} +roleRef: + apiGroup: rbac.authorization.k8s.io + kind: Role + name: {{ include "hami-vgpu.scheduler" . }} +subjects: + - kind: ServiceAccount + name: {{ include "hami-vgpu.scheduler" . }} + namespace: {{ include "hami-vgpu.namespace" . }} +--- +{{- if .Values.mockDevicePlugin.enabled }} +apiVersion: rbac.authorization.k8s.io/v1 +kind: ClusterRoleBinding +metadata: + name: {{ include "hami-vgpu.mock-device-plugin" . }} +roleRef: + apiGroup: rbac.authorization.k8s.io + kind: ClusterRole + name: {{ include "hami-vgpu.mock-device-plugin" . }} +subjects: + - kind: ServiceAccount + name: {{ include "hami-vgpu.mock-device-plugin" . }} + namespace: {{ include "hami-vgpu.namespace" . }} +{{- end -}} diff --git a/packages/system/hami/charts/hami/templates/scheduler/service.yaml b/packages/system/hami/charts/hami/templates/scheduler/service.yaml new file mode 100644 index 00000000..d7538fed --- /dev/null +++ b/packages/system/hami/charts/hami/templates/scheduler/service.yaml @@ -0,0 +1,34 @@ +apiVersion: v1 +kind: Service +metadata: + name: {{ include "hami-vgpu.scheduler" . }} + namespace: {{ include "hami-vgpu.namespace" . }} + labels: + app.kubernetes.io/component: hami-scheduler + {{- include "hami-vgpu.labels" . | nindent 4 }} + {{- if .Values.scheduler.service.labels }} + {{ toYaml .Values.scheduler.service.labels | indent 4 }} + {{- end }} + {{- if .Values.scheduler.service.annotations }} + annotations: {{ toYaml .Values.scheduler.service.annotations | nindent 4 }} + {{- end }} +spec: + type: {{ .Values.scheduler.service.type | default "NodePort" }} # Default type is NodePort + ports: + - name: http + port: {{ .Values.scheduler.service.httpPort | default 443 }} # Default HTTP port is 443 + targetPort: {{ .Values.scheduler.service.httpTargetPort | default 443 }} + {{- if eq (.Values.scheduler.service.type | default "NodePort") "NodePort" }} # If type is NodePort, set nodePort + nodePort: {{ .Values.scheduler.service.schedulerPort | default 31998 }} + {{- end }} + protocol: TCP + - name: monitor + port: {{ .Values.scheduler.service.monitorPort | default 31993 }} # Default monitoring port is 31993 + targetPort: {{ .Values.scheduler.service.monitorTargetPort | default 9395 }} + {{- if eq (.Values.scheduler.service.type | default "NodePort") "NodePort" }} # If type is NodePort, set nodePort + nodePort: {{ .Values.scheduler.service.monitorPort | default 31993 }} + {{- end }} + protocol: TCP + selector: + app.kubernetes.io/component: hami-scheduler + {{- include "hami-vgpu.selectorLabels" . | nindent 4 }} diff --git a/packages/system/hami/charts/hami/templates/scheduler/serviceaccount.yaml b/packages/system/hami/charts/hami/templates/scheduler/serviceaccount.yaml new file mode 100644 index 00000000..0435c003 --- /dev/null +++ b/packages/system/hami/charts/hami/templates/scheduler/serviceaccount.yaml @@ -0,0 +1,16 @@ +apiVersion: v1 +kind: ServiceAccount +metadata: + name: {{ include "hami-vgpu.scheduler" . }} + namespace: {{ include "hami-vgpu.namespace" . }} + labels: + app.kubernetes.io/component: "hami-scheduler" + {{- include "hami-vgpu.labels" . | nindent 4 }} +--- +{{- if .Values.mockDevicePlugin.enabled }} +apiVersion: v1 +kind: ServiceAccount +metadata: + name: {{ include "hami-vgpu.mock-device-plugin" . }} + namespace: {{ include "hami-vgpu.namespace" . }} +{{- end -}} diff --git a/packages/system/hami/charts/hami/templates/scheduler/webhook.yaml b/packages/system/hami/charts/hami/templates/scheduler/webhook.yaml new file mode 100644 index 00000000..db9f8029 --- /dev/null +++ b/packages/system/hami/charts/hami/templates/scheduler/webhook.yaml @@ -0,0 +1,57 @@ +{{- if .Values.scheduler.admissionWebhook.enabled -}} +apiVersion: admissionregistration.k8s.io/v1 +kind: MutatingWebhookConfiguration +metadata: + {{- if .Values.scheduler.certManager.enabled }} + annotations: + cert-manager.io/inject-ca-from: {{ include "hami-vgpu.namespace" . }}/{{ include "hami-vgpu.scheduler" . }}-serving-cert + {{- end }} + name: {{ include "hami-vgpu.scheduler.webhook" . }} +webhooks: + - admissionReviewVersions: + - v1beta1 + clientConfig: + {{- if .Values.scheduler.admissionWebhook.customURL.enabled }} + url: https://{{ .Values.scheduler.admissionWebhook.customURL.host}}:{{.Values.scheduler.admissionWebhook.customURL.port}}{{.Values.scheduler.admissionWebhook.customURL.path}} + {{- else }} + service: + name: {{ include "hami-vgpu.scheduler" . }} + namespace: {{ include "hami-vgpu.namespace" . }} + path: /webhook + port: {{ .Values.scheduler.service.httpPort }} + {{- end }} + failurePolicy: {{ .Values.scheduler.admissionWebhook.failurePolicy }} + matchPolicy: Equivalent + name: vgpu.hami.io + namespaceSelector: + matchExpressions: + - key: hami.io/webhook + operator: NotIn + values: + - ignore + {{- if .Values.scheduler.admissionWebhook.whitelistNamespaces }} + - key: kubernetes.io/metadata.name + operator: NotIn + values: + {{- toYaml .Values.scheduler.admissionWebhook.whitelistNamespaces | nindent 10 }} + {{- end }} + objectSelector: + matchExpressions: + - key: hami.io/webhook + operator: NotIn + values: + - ignore + reinvocationPolicy: {{ .Values.scheduler.admissionWebhook.reinvocationPolicy }} + rules: + - apiGroups: + - "" + apiVersions: + - v1 + operations: + - CREATE + resources: + - pods + scope: '*' + sideEffects: None + timeoutSeconds: 10 +{{- end }} diff --git a/packages/system/hami/charts/hami/values.yaml b/packages/system/hami/charts/hami/values.yaml new file mode 100644 index 00000000..01caca0b --- /dev/null +++ b/packages/system/hami/charts/hami/values.yaml @@ -0,0 +1,455 @@ +## @section Global configuration +## @param global.imageRegistry Global Docker image registry +## @param global.imagePullSecrets Global Docker image pull secrets +global: + ## @param global.imageRegistry Global Docker image registry + imageRegistry: "" + ## E.g. + ## imagePullSecrets: + ## - myRegistryKeySecretName + ## @param global.imagePullSecrets Global Docker image pull secrets + imagePullSecrets: [] + imageTag: "v2.8.1" + gpuHookPath: /usr/local + labels: {} + annotations: {} + managedNodeSelectorEnable: false + managedNodeSelector: + usage: "gpu" + +nameOverride: "" +fullnameOverride: "" +namespaceOverride: "" + +#Nvidia GPU Parameters +resourceName: "nvidia.com/gpu" +resourceMem: "nvidia.com/gpumem" +resourceMemPercentage: "nvidia.com/gpumem-percentage" +resourceCores: "nvidia.com/gpucores" +resourcePriority: "nvidia.com/priority" + +#MLU Parameters +mluResourceName: "cambricon.com/vmlu" +mluResourceMem: "cambricon.com/mlu.smlu.vmemory" +mluResourceCores: "cambricon.com/mlu.smlu.vcore" + +#Hygon DCU Parameters +dcuResourceName: "hygon.com/dcunum" +dcuResourceMem: "hygon.com/dcumem" +dcuResourceCores: "hygon.com/dcucores" + +#Metax sGPU Parameters +metaxResourceName: "metax-tech.com/sgpu" +metaxResourceCore: "metax-tech.com/vcore" +metaxResourceMem: "metax-tech.com/vmemory" +metaxsGPUTopologyAware: "false" + +#Enflame VGCU Parameters +enflameResourceNameVGCU: "enflame.com/vgcu" +enflameResourceNameVGCUPercentage: "enflame.com/vgcu-percentage" + +#Kunlun XPU Parameters +kunlunResourceName: "kunlunxin.com/xpu" +kunlunResourceVCountName: "kunlunxin.com/vxpu" +kunlunResourceVMemoryName: "kunlunxin.com/vxpu-memory" + +schedulerName: "hami-scheduler" + +scheduler: + # @param nodeName defines the node name and the nvidia-vgpu-scheduler-scheduler will schedule to the node. + # if we install the nvidia-vgpu-scheduler-scheduler as default scheduler, we need to remove the k8s default + # scheduler pod from the cluster first, we must specify node name to skip the schedule workflow. + nodeName: "" + #nodeLabelSelector: + # "gpu": "on" + overwriteEnv: "false" + defaultSchedulerPolicy: + nodeSchedulerPolicy: binpack + gpuSchedulerPolicy: spread + metricsBindAddress: ":9395" + # If set to false, When Pod.Spec.SchedulerName equals to the const DefaultSchedulerName in k8s.io/api/core/v1 package, webhook will not overwrite it + forceOverwriteDefaultScheduler: true + livenessProbe: false + leaderElect: true + # when leaderElect is true, replicas is available, otherwise replicas is 1. + replicas: 1 + kubeScheduler: + # @param enabled indicate whether to run kube-scheduler container in the scheduler pod, it's true by default. + enabled: true + ## @param image.registry kube scheduler image registry + ## @param image.repository kube scheduler image repository + ## @param image.tag kube scheduler image tag (immutable tags are recommended) + ## @param image.pullPolicy kube scheduler image pull policy + ## @param image.pullSecrets Specify docker-registry secret names as an array + image: + registry: "registry.cn-hangzhou.aliyuncs.com" + repository: "google_containers/kube-scheduler" + tag: "" + ## Specify a imagePullPolicy + ## Defaults to 'Always' if image tag is 'latest', else set to 'IfNotPresent' + ## ref: https://kubernetes.io/docs/user-guide/images/#pre-pulling-images + ## + pullPolicy: IfNotPresent + ## Optionally specify an array of imagePullSecrets. + ## Secrets must be manually created in the namespace. + ## Example: + ## pullSecrets: + ## - myRegistryKeySecretName + ## + pullSecrets: [] + resources: {} + # If you do want to specify resources, uncomment the following lines, adjust them as necessary. + # and remove the curly braces after 'resources:'. +# limits: +# cpu: 1000m +# memory: 1000Mi +# requests: +# cpu: 100m +# memory: 100Mi + extraNewArgs: + - --config=/config/config.yaml + - -v=4 + extraArgs: + - --policy-config-file=/config/config.json + - -v=4 + extender: + ## @param image.registry scheduler extender image registry + ## @param image.repository scheduler extender image repository + ## @param image.tag scheduler extender image tag (immutable tags are recommended) + ## @param image.pullPolicy scheduler extender image pull policy + ## @param image.pullSecrets Specify docker-registry secret names as an array + image: + registry: "docker.io" + repository: "projecthami/hami" + tag: "" + ## Specify a imagePullPolicy + ## Defaults to 'Always' if image tag is 'latest', else set to 'IfNotPresent' + ## ref: https://kubernetes.io/docs/user-guide/images/#pre-pulling-images + ## + pullPolicy: IfNotPresent + ## Optionally specify an array of imagePullSecrets. + ## Secrets must be manually created in the namespace. + ## Example: + ## pullSecrets: + ## - myRegistryKeySecretName + ## + pullSecrets: [] + resources: {} + # If you do want to specify resources, uncomment the following lines, adjust them as necessary, + # and remove the curly braces after 'resources:'. +# limits: +# cpu: 1000m +# memory: 1000Mi +# requests: +# cpu: 100m +# memory: 100Mi + extraArgs: + - --debug + - -v=4 + nodeLockExpire: "5m" + podAnnotations: {} + tolerations: [] + #serviceAccountName: "hami-vgpu-scheduler-sa" + admissionWebhook: + # If set to false, the admission webhook is not installed and any pods that should use HAMi must be + # configured to use the right 'schedulerName' and other device-specific configurations. + enabled: true + customURL: + enabled: false + # must be an endpoint using https. + # should generate host certs here + host: 127.0.0.1 # hostname or ip, can be your node'IP if you want to use https://:/ + port: 31998 + path: /webhook + whitelistNamespaces: + # Specify the namespaces that the webhook will not be applied to. + # - default + # - kube-system + # - istio-system + reinvocationPolicy: Never + failurePolicy: Ignore + ## TLS Certificate Option 1: Use cert-manager to generate self-signed certificate. + ## If enabled, always takes precedence over options 2. + certManager: + enabled: false + ## TLS Certificate Option 2: Use kube-webhook-certgen to generate self-signed certificate. + ## If true and certManager.enabled is false, Helm will automatically create a self-signed cert and secret for you. + patch: + enabled: true + ## @param image.registry scheduler extender image registry + ## @param image.repository scheduler extender image repository + ## @param image.tag scheduler extender image tag (immutable tags are recommended) + ## @param image.pullPolicy scheduler extender image pull policy + ## @param image.pullSecrets Specify docker-registry secret names as an array + image: + registry: "docker.io" + repository: "jettech/kube-webhook-certgen" + tag: "v1.5.2" + ## Specify a imagePullPolicy + ## Defaults to 'Always' if image tag is 'latest', else set to 'IfNotPresent' + ## ref: https://kubernetes.io/docs/user-guide/images/#pre-pulling-images + ## + pullPolicy: IfNotPresent + ## Optionally specify an array of imagePullSecrets. + ## Secrets must be manually created in the namespace. + ## Example: + ## pullSecrets: + ## - myRegistryKeySecretName + ## + pullSecrets: [] + ## @param image.registry scheduler extender image registry + ## @param image.repository scheduler extender image repository + ## @param image.tag scheduler extender image tag (immutable tags are recommended) + ## @param image.pullPolicy scheduler extender image pull policy + ## @param image.pullSecrets Specify docker-registry secret names as an array + imageNew: + registry: "docker.io" + repository: "liangjw/kube-webhook-certgen" + tag: "v1.1.1" + ## Specify a imagePullPolicy + ## Defaults to 'Always' if image tag is 'latest', else set to 'IfNotPresent' + ## ref: https://kubernetes.io/docs/user-guide/images/#pre-pulling-images + ## + pullPolicy: IfNotPresent + ## Optionally specify an array of imagePullSecrets. + ## Secrets must be manually created in the namespace. + ## Example: + ## pullSecrets: + ## - myRegistryKeySecretName + ## + pullSecrets: [] + priorityClassName: "" + podAnnotations: {} + nodeSelector: {} + tolerations: [] + runAsUser: 2000 + service: + type: NodePort # Default type is NodePort, can be changed to ClusterIP + httpPort: 443 # HTTP port + schedulerPort: 31998 # NodePort for HTTP + monitorPort: 31993 # Monitoring port + monitorTargetPort: 9395 + httpTargetPort: 443 + labels: {} + annotations: {} + +devicePlugin: + enabled: true + gpuOperatorToolkitReady: + enabled: false + hostPath: "/run/nvidia/validations" + ## @param image.registry devicePlugin image registry + ## @param image.repository devicePlugin image repository + ## @param image.tag devicePlugin image tag (immutable tags are recommended) + ## @param image.pullPolicy devicePlugin image pull policy + ## @param image.pullSecrets Specify docker-registry secret names as an array + image: + registry: "docker.io" + repository: "projecthami/hami" + tag: "" + ## Specify a imagePullPolicy + ## Defaults to 'Always' if image tag is 'latest', else set to 'IfNotPresent' + ## ref: https://kubernetes.io/docs/user-guide/images/#pre-pulling-images + ## + pullPolicy: IfNotPresent + ## Optionally specify an array of imagePullSecrets. + ## Secrets must be manually created in the namespace. + ## Example: + ## pullSecrets: + ## - myRegistryKeySecretName + ## + pullSecrets: [] + monitor: + ## @param image.registry monitor image registry + ## @param image.repository monitor image repository + ## @param image.tag monitor image tag (immutable tags are recommended) + ## @param image.pullPolicy monitor image pull policy + ## @param image.pullSecrets Specify docker-registry secret names as an array + image: + registry: "docker.io" + repository: "projecthami/hami" + tag: "" + ## Specify a imagePullPolicy + ## Defaults to 'Always' if image tag is 'latest', else set to 'IfNotPresent' + ## ref: https://kubernetes.io/docs/user-guide/images/#pre-pulling-images + ## + pullPolicy: IfNotPresent + ## Optionally specify an array of imagePullSecrets. + ## Secrets must be manually created in the namespace. + ## Example: + ## pullSecrets: + ## - myRegistryKeySecretName + ## + pullSecrets: [] + ctrPath: /usr/local/vgpu/containers + resyncInterval: "5m" + extraArgs: + - -v=4 + extraEnvs: {} + resources: {} + # If you do want to specify resources, uncomment the following lines, adjust them as necessary. + # and remove the curly braces after 'resources:'. +# limits: +# cpu: 1000m +# memory: 1000Mi +# requests: +# cpu: 100m +# memory: 100Mi + + deviceSplitCount: 10 + deviceMemoryScaling: 1 + deviceCoreScaling: 1 + # Node configuration for device plugin, Priority: externalConfigName > config > default config + nodeConfiguration: + # If you want to use a custom config.json, you can set the content here. + # If this is set, it will override the default config.json(An example is as follows). + config: | + { + "nodeconfig": [ + { + "name": "your-node-name", + "operatingmode": "hami-core", + "devicememoryscaling": 1, + "devicesplitcount": 10, + "migstrategy": "none", + "filterdevices": { + "uuid": [], + "index": [] + } + } + ] + } + # If you want to use an existing ConfigMap, you can set the name here. + # If this is set, the chart will not create the ConfigMap and will use the existing one. + externalConfigName: "" + # The runtime class name to be used by the device plugin, and added to the pod.spec.runtimeClassName of applications utilizing NVIDIA GPUs + runtimeClassName: "" + # Whether to create runtime class, name comes from runtimeClassName when it is set + createRuntimeClass: false + migStrategy: "none" + disablecorelimit: "false" + passDeviceSpecsEnabled: false + deviceListStrategy: "envvar" + nvidiaHookPath: null + nvidiaDriverRoot: null + gdrcopyEnabled: null + gdsEnabled: null + mofedEnabled: null + + extraArgs: + - -v=4 + extraEnvs: {} + service: + type: NodePort # Default type is NodePort, can be changed to ClusterIP + httpPort: 31992 + labels: {} + annotations: {} + + pluginPath: /var/lib/kubelet/device-plugins + libPath: /usr/local/vgpu + + podAnnotations: {} + nvidiaNodeSelector: + gpu: "on" + tolerations: [] + # The updateStrategy for DevicePlugin DaemonSet. + # If you want to update the DaemonSet by manual, set type as "OnDelete". + # We recommend use OnDelete update strategy because DevicePlugin pod restart will cause business pod restart, this behavior is destructive. + # Otherwise, you can use RollingUpdate update strategy to rolling update DevicePlugin pod. + updateStrategy: + type: RollingUpdate + rollingUpdate: + maxUnavailable: 1 + + resources: {} + # If you do want to specify resources, uncomment the following lines, adjust them as necessary. + # and remove the curly braces after 'resources:'. +# limits: +# cpu: 1000m +# memory: 1000Mi +# requests: +# cpu: 100m +# memory: 100Mi + +mockDevicePlugin: + enabled: false + image: + registry: "docker.io" + repository: "projecthami/mock-device-plugin" + tag: "1.0.1" + ## Specify a imagePullPolicy + ## Defaults to 'Always' if image tag is 'latest', else set to 'IfNotPresent' + ## ref: https://kubernetes.io/docs/user-guide/images/#pre-pulling-images + ## + pullPolicy: IfNotPresent + ## Optionally specify an array of imagePullSecrets. + ## Secrets must be manually created in the namespace. + ## Example: + ## pullSecrets: + ## - myRegistryKeySecretName + ## + pullSecrets: [] +devices: + amd: + customresources: + - amd.com/gpu + - amd.com/gpu-memory + awsneuron: + customresources: + - aws.amazon.com/neuron + - aws.amazon.com/neuroncore + kunlun: + enabled: true + customresources: + - kunlunxin.com/xpu + - kunlunxin.com/vxpu + - kunlunxin.com/vxpu-memory + enflame: + enabled: true + customresources: + - enflame.com/vgcu + - enflame.com/vgcu-percentage + - enflame.com/gcu + mthreads: + enabled: true + customresources: + - mthreads.com/vgpu + nvidia: + gpuCorePolicy: default + libCudaLogLevel: 1 + ascend: + enabled: false + image: "" + imagePullPolicy: IfNotPresent + extraArgs: [] + nodeSelector: + ascend: "on" + tolerations: [] + customresources: + - huawei.com/Ascend910A + - huawei.com/Ascend910A-memory + - huawei.com/Ascend910B2 + - huawei.com/Ascend910B2-memory + - huawei.com/Ascend910B3 + - huawei.com/Ascend910B3-memory + - huawei.com/Ascend910B4 + - huawei.com/Ascend910B4-memory + - huawei.com/Ascend910B4-1 + - huawei.com/Ascend910B4-1-memory + - huawei.com/Ascend310P + - huawei.com/Ascend310P-memory + iluvatar: + enabled: false + customresources: + - iluvatar.ai/BI-V100-vgpu + - iluvatar.ai/BI-V100.vCore + - iluvatar.ai/BI-V100.vMem + - iluvatar.ai/BI-V150-vgpu + - iluvatar.ai/BI-V150.vCore + - iluvatar.ai/BI-V150.vMem + - iluvatar.ai/MR-V100-vgpu + - iluvatar.ai/MR-V100.vCore + - iluvatar.ai/MR-V100.vMem + - iluvatar.ai/MR-V50-vgpu + - iluvatar.ai/MR-V50.vCore + - iluvatar.ai/MR-V50.vMem diff --git a/packages/system/hami/values.yaml b/packages/system/hami/values.yaml new file mode 100644 index 00000000..76e8d631 --- /dev/null +++ b/packages/system/hami/values.yaml @@ -0,0 +1,17 @@ +hami: + scheduler: + kubeScheduler: + image: + registry: registry.k8s.io + repository: kube-scheduler + devicePlugin: + runtimeClassName: nvidia + updateStrategy: + type: RollingUpdate + # See https://github.com/Project-HAMi/HAMi/blob/v2.8.1/deployments/charts/hami/values.yaml + # Each entry: {"name": "node-name", "operatingmode": "hami-core", "devicesplitcount": N, ...} + nodeConfiguration: + config: | + { + "nodeconfig": [] + } diff --git a/packages/system/kubernetes-rd/cozyrds/kubernetes.yaml b/packages/system/kubernetes-rd/cozyrds/kubernetes.yaml index 1631c86e..f81eaff0 100644 --- a/packages/system/kubernetes-rd/cozyrds/kubernetes.yaml +++ b/packages/system/kubernetes-rd/cozyrds/kubernetes.yaml @@ -22,7 +22,7 @@ spec: singular: kubernetes plural: kuberneteses openAPISchema: |- - {"title":"Chart Values","type":"object","properties":{"storageClass":{"description":"StorageClass used to store the data.","type":"string","default":"replicated"},"nodeGroups":{"description":"Worker nodes configuration map.","type":"object","default":{"md0":{"ephemeralStorage":"20Gi","gpus":[],"instanceType":"u1.medium","maxReplicas":10,"minReplicas":0,"resources":{},"roles":["ingress-nginx"]}},"additionalProperties":{"type":"object","required":["ephemeralStorage","instanceType","maxReplicas","minReplicas","resources"],"properties":{"ephemeralStorage":{"description":"Ephemeral storage size.","default":"20Gi","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true},"gpus":{"description":"List of GPUs to attach (NVIDIA driver requires at least 4 GiB RAM).","type":"array","items":{"type":"object","required":["name"],"properties":{"name":{"description":"Name of GPU, such as \"nvidia.com/AD102GL_L40S\".","type":"string"}}}},"instanceType":{"description":"Virtual machine instance type.","type":"string","default":"u1.medium"},"maxReplicas":{"description":"Maximum number of replicas.","type":"integer","default":10},"minReplicas":{"description":"Minimum number of replicas.","type":"integer","default":0},"resources":{"description":"CPU and memory resources for each worker node.","type":"object","properties":{"cpu":{"description":"CPU available.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true},"memory":{"description":"Memory (RAM) available.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true}}},"roles":{"description":"List of node roles.","type":"array","items":{"type":"string"}}}}},"version":{"description":"Kubernetes major.minor version to deploy","type":"string","default":"v1.35","enum":["v1.35","v1.34","v1.33","v1.32","v1.31","v1.30"]},"host":{"description":"External hostname for Kubernetes cluster. Defaults to `.` if empty.","type":"string","default":""},"addons":{"description":"Cluster addons configuration.","type":"object","default":{},"required":["certManager","cilium","coredns","fluxcd","gatewayAPI","gpuOperator","ingressNginx","monitoringAgents","velero","verticalPodAutoscaler"],"properties":{"certManager":{"description":"Cert-manager addon.","type":"object","default":{},"required":["enabled","valuesOverride"],"properties":{"enabled":{"description":"Enable cert-manager.","type":"boolean","default":false},"valuesOverride":{"description":"Custom Helm values overrides.","type":"object","default":{},"x-kubernetes-preserve-unknown-fields":true}}},"cilium":{"description":"Cilium CNI plugin.","type":"object","default":{},"required":["valuesOverride"],"properties":{"valuesOverride":{"description":"Custom Helm values overrides.","type":"object","default":{},"x-kubernetes-preserve-unknown-fields":true}}},"coredns":{"description":"CoreDNS addon.","type":"object","default":{},"required":["valuesOverride"],"properties":{"valuesOverride":{"description":"Custom Helm values overrides.","type":"object","default":{},"x-kubernetes-preserve-unknown-fields":true}}},"fluxcd":{"description":"FluxCD GitOps operator.","type":"object","default":{},"required":["enabled","valuesOverride"],"properties":{"enabled":{"description":"Enable FluxCD.","type":"boolean","default":false},"valuesOverride":{"description":"Custom Helm values overrides.","type":"object","default":{},"x-kubernetes-preserve-unknown-fields":true}}},"gatewayAPI":{"description":"Gateway API addon.","type":"object","default":{},"required":["enabled"],"properties":{"enabled":{"description":"Enable Gateway API.","type":"boolean","default":false}}},"gpuOperator":{"description":"NVIDIA GPU Operator.","type":"object","default":{},"required":["enabled","valuesOverride"],"properties":{"enabled":{"description":"Enable GPU Operator.","type":"boolean","default":false},"valuesOverride":{"description":"Custom Helm values overrides.","type":"object","default":{},"x-kubernetes-preserve-unknown-fields":true}}},"ingressNginx":{"description":"Ingress-NGINX controller.","type":"object","default":{},"required":["enabled","exposeMethod","valuesOverride"],"properties":{"enabled":{"description":"Enable the controller (requires nodes labeled `ingress-nginx`).","type":"boolean","default":false},"exposeMethod":{"description":"Method to expose the controller. Allowed values: `Proxied`, `LoadBalancer`.","type":"string","default":"Proxied","enum":["Proxied","LoadBalancer"]},"hosts":{"description":"Domains routed to this tenant cluster when `exposeMethod` is `Proxied`.","type":"array","default":[],"items":{"type":"string"}},"valuesOverride":{"description":"Custom Helm values overrides.","type":"object","default":{},"x-kubernetes-preserve-unknown-fields":true}}},"monitoringAgents":{"description":"Monitoring agents.","type":"object","default":{},"required":["enabled","valuesOverride"],"properties":{"enabled":{"description":"Enable monitoring agents.","type":"boolean","default":false},"valuesOverride":{"description":"Custom Helm values overrides.","type":"object","default":{},"x-kubernetes-preserve-unknown-fields":true}}},"velero":{"description":"Velero backup/restore addon.","type":"object","default":{},"required":["enabled","valuesOverride"],"properties":{"enabled":{"description":"Enable Velero.","type":"boolean","default":false},"valuesOverride":{"description":"Custom Helm values overrides.","type":"object","default":{},"x-kubernetes-preserve-unknown-fields":true}}},"verticalPodAutoscaler":{"description":"Vertical Pod Autoscaler.","type":"object","default":{},"required":["valuesOverride"],"properties":{"valuesOverride":{"description":"Custom Helm values overrides.","type":"object","default":{},"x-kubernetes-preserve-unknown-fields":true}}}}},"controlPlane":{"description":"Kubernetes control-plane configuration.","type":"object","default":{},"required":["apiServer","controllerManager","konnectivity","replicas","scheduler"],"properties":{"apiServer":{"description":"API Server configuration.","type":"object","default":{},"required":["resources","resourcesPreset"],"properties":{"resources":{"description":"CPU and memory resources for API Server.","type":"object","default":{},"properties":{"cpu":{"description":"CPU available.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true},"memory":{"description":"Memory (RAM) available.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true}}},"resourcesPreset":{"description":"Preset if `resources` omitted.","type":"string","default":"large","enum":["nano","micro","small","medium","large","xlarge","2xlarge"]}}},"controllerManager":{"description":"Controller Manager configuration.","type":"object","default":{},"required":["resources","resourcesPreset"],"properties":{"resources":{"description":"CPU and memory resources for Controller Manager.","type":"object","default":{},"properties":{"cpu":{"description":"CPU available.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true},"memory":{"description":"Memory (RAM) available.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true}}},"resourcesPreset":{"description":"Preset if `resources` omitted.","type":"string","default":"micro","enum":["nano","micro","small","medium","large","xlarge","2xlarge"]}}},"konnectivity":{"description":"Konnectivity configuration.","type":"object","default":{},"required":["server"],"properties":{"server":{"description":"Konnectivity Server configuration.","type":"object","default":{},"required":["resources","resourcesPreset"],"properties":{"resources":{"description":"CPU and memory resources for Konnectivity.","type":"object","default":{},"properties":{"cpu":{"description":"CPU available.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true},"memory":{"description":"Memory (RAM) available.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true}}},"resourcesPreset":{"description":"Preset if `resources` omitted.","type":"string","default":"micro","enum":["nano","micro","small","medium","large","xlarge","2xlarge"]}}}}},"replicas":{"description":"Number of control-plane replicas.","type":"integer","default":2},"scheduler":{"description":"Scheduler configuration.","type":"object","default":{},"required":["resources","resourcesPreset"],"properties":{"resources":{"description":"CPU and memory resources for Scheduler.","type":"object","default":{},"properties":{"cpu":{"description":"CPU available.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true},"memory":{"description":"Memory (RAM) available.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true}}},"resourcesPreset":{"description":"Preset if `resources` omitted.","type":"string","default":"micro","enum":["nano","micro","small","medium","large","xlarge","2xlarge"]}}}}},"images":{"description":"Optional image overrides for air-gapped or rate-limited registries.","type":"object","default":{},"properties":{"waitForKubeconfig":{"description":"Image used by the wait-for-kubeconfig init container. Empty falls back to images/busybox.tag.","type":"string","default":""}}}}} + {"title":"Chart Values","type":"object","properties":{"storageClass":{"description":"StorageClass used to store the data.","type":"string","default":"replicated"},"nodeGroups":{"description":"Worker nodes configuration map.","type":"object","default":{"md0":{"ephemeralStorage":"20Gi","gpus":[],"instanceType":"u1.medium","maxReplicas":10,"minReplicas":0,"resources":{},"roles":["ingress-nginx"]}},"additionalProperties":{"type":"object","required":["ephemeralStorage","instanceType","maxReplicas","minReplicas","resources"],"properties":{"ephemeralStorage":{"description":"Ephemeral storage size.","default":"20Gi","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true},"gpus":{"description":"List of GPUs to attach (NVIDIA driver requires at least 4 GiB RAM).","type":"array","items":{"type":"object","required":["name"],"properties":{"name":{"description":"Name of GPU, such as \"nvidia.com/AD102GL_L40S\".","type":"string"}}}},"instanceType":{"description":"Virtual machine instance type.","type":"string","default":"u1.medium"},"maxReplicas":{"description":"Maximum number of replicas.","type":"integer","default":10},"minReplicas":{"description":"Minimum number of replicas.","type":"integer","default":0},"resources":{"description":"CPU and memory resources for each worker node.","type":"object","properties":{"cpu":{"description":"CPU available.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true},"memory":{"description":"Memory (RAM) available.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true}}},"roles":{"description":"List of node roles.","type":"array","items":{"type":"string"}}}}},"version":{"description":"Kubernetes major.minor version to deploy","type":"string","default":"v1.35","enum":["v1.35","v1.34","v1.33","v1.32","v1.31","v1.30"]},"host":{"description":"External hostname for Kubernetes cluster. Defaults to `.` if empty.","type":"string","default":""},"addons":{"description":"Cluster addons configuration.","type":"object","default":{},"required":["certManager","cilium","coredns","fluxcd","gatewayAPI","gpuOperator","hami","ingressNginx","monitoringAgents","velero","verticalPodAutoscaler"],"properties":{"certManager":{"description":"Cert-manager addon.","type":"object","default":{},"required":["enabled","valuesOverride"],"properties":{"enabled":{"description":"Enable cert-manager.","type":"boolean","default":false},"valuesOverride":{"description":"Custom Helm values overrides.","type":"object","default":{},"x-kubernetes-preserve-unknown-fields":true}}},"cilium":{"description":"Cilium CNI plugin.","type":"object","default":{},"required":["valuesOverride"],"properties":{"valuesOverride":{"description":"Custom Helm values overrides.","type":"object","default":{},"x-kubernetes-preserve-unknown-fields":true}}},"coredns":{"description":"CoreDNS addon.","type":"object","default":{},"required":["valuesOverride"],"properties":{"valuesOverride":{"description":"Custom Helm values overrides.","type":"object","default":{},"x-kubernetes-preserve-unknown-fields":true}}},"fluxcd":{"description":"FluxCD GitOps operator.","type":"object","default":{},"required":["enabled","valuesOverride"],"properties":{"enabled":{"description":"Enable FluxCD.","type":"boolean","default":false},"valuesOverride":{"description":"Custom Helm values overrides.","type":"object","default":{},"x-kubernetes-preserve-unknown-fields":true}}},"gatewayAPI":{"description":"Gateway API addon.","type":"object","default":{},"required":["enabled"],"properties":{"enabled":{"description":"Enable Gateway API.","type":"boolean","default":false}}},"gpuOperator":{"description":"NVIDIA GPU Operator.","type":"object","default":{},"required":["enabled","valuesOverride"],"properties":{"enabled":{"description":"Enable GPU Operator.","type":"boolean","default":false},"valuesOverride":{"description":"Custom Helm values overrides.","type":"object","default":{},"x-kubernetes-preserve-unknown-fields":true}}},"hami":{"description":"HAMi GPU virtualization middleware.","type":"object","default":{},"required":["enabled","valuesOverride"],"properties":{"enabled":{"description":"Enable HAMi (requires GPU Operator).","type":"boolean","default":false},"valuesOverride":{"description":"Custom Helm values overrides.","type":"object","default":{},"x-kubernetes-preserve-unknown-fields":true}}},"ingressNginx":{"description":"Ingress-NGINX controller.","type":"object","default":{},"required":["enabled","exposeMethod","valuesOverride"],"properties":{"enabled":{"description":"Enable the controller (requires nodes labeled `ingress-nginx`).","type":"boolean","default":false},"exposeMethod":{"description":"Method to expose the controller. Allowed values: `Proxied`, `LoadBalancer`.","type":"string","default":"Proxied","enum":["Proxied","LoadBalancer"]},"hosts":{"description":"Domains routed to this tenant cluster when `exposeMethod` is `Proxied`.","type":"array","default":[],"items":{"type":"string"}},"valuesOverride":{"description":"Custom Helm values overrides.","type":"object","default":{},"x-kubernetes-preserve-unknown-fields":true}}},"monitoringAgents":{"description":"Monitoring agents.","type":"object","default":{},"required":["enabled","valuesOverride"],"properties":{"enabled":{"description":"Enable monitoring agents.","type":"boolean","default":false},"valuesOverride":{"description":"Custom Helm values overrides.","type":"object","default":{},"x-kubernetes-preserve-unknown-fields":true}}},"velero":{"description":"Velero backup/restore addon.","type":"object","default":{},"required":["enabled","valuesOverride"],"properties":{"enabled":{"description":"Enable Velero.","type":"boolean","default":false},"valuesOverride":{"description":"Custom Helm values overrides.","type":"object","default":{},"x-kubernetes-preserve-unknown-fields":true}}},"verticalPodAutoscaler":{"description":"Vertical Pod Autoscaler.","type":"object","default":{},"required":["valuesOverride"],"properties":{"valuesOverride":{"description":"Custom Helm values overrides.","type":"object","default":{},"x-kubernetes-preserve-unknown-fields":true}}}}},"controlPlane":{"description":"Kubernetes control-plane configuration.","type":"object","default":{},"required":["apiServer","controllerManager","konnectivity","replicas","scheduler"],"properties":{"apiServer":{"description":"API Server configuration.","type":"object","default":{},"required":["resources","resourcesPreset"],"properties":{"resources":{"description":"CPU and memory resources for API Server.","type":"object","default":{},"properties":{"cpu":{"description":"CPU available.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true},"memory":{"description":"Memory (RAM) available.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true}}},"resourcesPreset":{"description":"Preset if `resources` omitted.","type":"string","default":"large","enum":["nano","micro","small","medium","large","xlarge","2xlarge"]}}},"controllerManager":{"description":"Controller Manager configuration.","type":"object","default":{},"required":["resources","resourcesPreset"],"properties":{"resources":{"description":"CPU and memory resources for Controller Manager.","type":"object","default":{},"properties":{"cpu":{"description":"CPU available.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true},"memory":{"description":"Memory (RAM) available.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true}}},"resourcesPreset":{"description":"Preset if `resources` omitted.","type":"string","default":"micro","enum":["nano","micro","small","medium","large","xlarge","2xlarge"]}}},"konnectivity":{"description":"Konnectivity configuration.","type":"object","default":{},"required":["server"],"properties":{"server":{"description":"Konnectivity Server configuration.","type":"object","default":{},"required":["resources","resourcesPreset"],"properties":{"resources":{"description":"CPU and memory resources for Konnectivity.","type":"object","default":{},"properties":{"cpu":{"description":"CPU available.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true},"memory":{"description":"Memory (RAM) available.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true}}},"resourcesPreset":{"description":"Preset if `resources` omitted.","type":"string","default":"micro","enum":["nano","micro","small","medium","large","xlarge","2xlarge"]}}}}},"replicas":{"description":"Number of control-plane replicas.","type":"integer","default":2},"scheduler":{"description":"Scheduler configuration.","type":"object","default":{},"required":["resources","resourcesPreset"],"properties":{"resources":{"description":"CPU and memory resources for Scheduler.","type":"object","default":{},"properties":{"cpu":{"description":"CPU available.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true},"memory":{"description":"Memory (RAM) available.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true}}},"resourcesPreset":{"description":"Preset if `resources` omitted.","type":"string","default":"micro","enum":["nano","micro","small","medium","large","xlarge","2xlarge"]}}}}},"images":{"description":"Optional image overrides for air-gapped or rate-limited registries.","type":"object","default":{},"properties":{"waitForKubeconfig":{"description":"Image used by the wait-for-kubeconfig init container. Empty falls back to images/busybox.tag.","type":"string","default":""}}}}} release: prefix: kubernetes- labels: @@ -40,7 +40,7 @@ spec: tags: - compute icon: <svg width="144" height="144" viewBox="0 0 144 144" fill="none" xmlns="http://www.w3.org/2000/svg">
<rect width="144" height="144" rx="24" fill="url(#paint0_linear_681_2845)"/>
<path d="M71.9968 19C70.3039 19.0002 68.9312 20.5322 68.9314 22.4221C68.9314 22.4511 68.9373 22.4788 68.9379 22.5076C68.9354 22.7644 68.9231 23.0737 68.9314 23.2973C68.9717 24.3873 69.2082 25.2216 69.3506 26.2258C69.6084 28.3752 69.8245 30.1569 69.6912 31.813C69.5615 32.4375 69.1037 33.0086 68.6956 33.4056L68.6235 34.7086C66.7839 34.8617 64.9319 35.1421 63.0821 35.5641C55.1226 37.3798 48.2695 41.4991 43.052 47.0609C42.7134 46.8288 42.1211 46.4019 41.945 46.2712C41.3977 46.3454 40.8445 46.5151 40.1241 46.0935C38.7522 45.1657 37.5028 43.8851 35.991 42.3424C35.2982 41.6044 34.7966 40.9018 33.9735 40.1904C33.7866 40.0289 33.5014 39.8104 33.2923 39.6442C32.6489 39.1288 31.89 38.86 31.157 38.8348C30.2147 38.8024 29.3075 39.1725 28.7138 39.9206C27.6584 41.2506 27.9963 43.2833 29.4671 44.4614C29.482 44.4734 29.4979 44.4827 29.5129 44.4943C29.715 44.6589 29.9625 44.8698 30.1483 45.0076C31.0217 45.6555 31.8195 45.9872 32.6897 46.5015C34.5231 47.6391 36.043 48.5823 37.2486 49.7196C37.7194 50.2237 37.8016 51.1122 37.8643 51.4964L38.8468 52.3782C33.5872 60.3308 31.153 70.1539 32.5915 80.1627L31.3077 80.5378C30.9693 80.9768 30.4912 81.6676 29.9911 81.8738C28.4138 82.3729 26.6387 82.5562 24.4956 82.7819C23.4894 82.866 22.6213 82.8158 21.5546 83.0188C21.3198 83.0635 20.9927 83.1491 20.7358 83.2097C20.7269 83.2116 20.7186 83.2142 20.7096 83.2162C20.6956 83.2195 20.6772 83.2263 20.6638 83.2294C18.857 83.668 17.6963 85.3365 18.0699 86.9805C18.4437 88.6248 20.2086 89.6248 22.0262 89.2312C22.0393 89.2282 22.0584 89.2277 22.072 89.2246C22.0926 89.2199 22.1106 89.2099 22.131 89.2049C22.3844 89.149 22.7019 89.0868 22.9236 89.0272C23.9723 88.7451 24.7318 88.3306 25.6746 87.9677C27.7029 87.2368 29.3828 86.6262 31.0195 86.3883C31.703 86.3345 32.4232 86.812 32.7814 87.0134L34.1177 86.7831C37.1926 96.3613 43.6366 104.103 51.7963 108.961L51.2396 110.303C51.4403 110.824 51.6616 111.53 51.5121 112.045C50.9171 113.595 49.898 115.231 48.7374 117.055C48.1755 117.898 47.6004 118.552 47.0934 119.516C46.972 119.747 46.8175 120.102 46.7004 120.346C45.9125 122.039 46.4904 123.99 48.0038 124.722C49.5268 125.459 51.4171 124.682 52.2352 122.985C52.2364 122.982 52.2406 122.98 52.2417 122.978C52.2426 122.976 52.2409 122.973 52.2417 122.971C52.3582 122.731 52.5233 122.415 52.6216 122.188C53.056 121.189 53.2005 120.332 53.5059 119.365C54.317 117.318 54.7626 115.17 55.8791 113.832C56.1849 113.466 56.6833 113.325 57.2001 113.186L57.8944 111.922C65.008 114.665 72.9705 115.402 80.9245 113.587C82.7391 113.173 84.4908 112.637 86.1843 111.994C86.3794 112.342 86.742 113.011 86.8393 113.179C87.3644 113.351 87.9377 113.439 88.4047 114.133C89.2401 115.567 89.8114 117.263 90.5073 119.312C90.8128 120.279 90.9638 121.136 91.3981 122.136C91.4971 122.363 91.6614 122.684 91.778 122.925C92.5944 124.628 94.4907 125.407 96.0159 124.669C97.5292 123.937 98.1077 121.986 97.3194 120.293C97.2023 120.049 97.0412 119.695 96.9198 119.464C96.4127 118.499 95.8377 117.852 95.2758 117.009C94.1152 115.185 93.1526 113.67 92.5575 112.12C92.3087 111.32 92.5995 110.823 92.7933 110.303C92.6772 110.17 92.4288 109.414 92.2824 109.059C100.762 104.029 107.017 95.9985 109.955 86.7239C110.351 86.7865 111.041 86.9091 111.265 86.9542C111.726 86.6487 112.149 86.2501 112.981 86.3159C114.617 86.5537 116.297 87.1645 118.326 87.8953C119.268 88.2581 120.028 88.6793 121.077 88.9614C121.298 89.021 121.616 89.0766 121.869 89.1325C121.889 89.1375 121.908 89.1475 121.928 89.1522C121.942 89.1553 121.961 89.1558 121.974 89.1588C123.792 89.552 125.557 88.5526 125.93 86.9081C126.303 85.2641 125.143 83.5952 123.336 83.157C123.074 83.097 122.701 82.995 122.446 82.9465C121.379 82.7435 120.511 82.7935 119.505 82.7095C117.361 82.4839 115.586 82.3004 114.009 81.8014C113.366 81.5507 112.908 80.7819 112.686 80.4655L111.448 80.1035C112.09 75.438 111.917 70.5825 110.806 65.7243C109.685 60.8208 107.704 56.3361 105.062 52.3848C105.379 52.0948 105.979 51.5612 106.149 51.4043C106.199 50.8517 106.156 50.2722 106.725 49.6603C107.931 48.523 109.451 47.5799 111.284 46.4423C112.154 45.9279 112.959 45.5964 113.832 44.9484C114.03 44.8019 114.299 44.5699 114.507 44.4022C115.977 43.2237 116.316 41.1911 115.26 39.8614C114.204 38.5317 112.159 38.4065 110.688 39.585C110.479 39.7516 110.195 39.9688 110.007 40.1312C109.184 40.8426 108.676 41.5452 107.983 42.2832C106.471 43.8259 105.222 45.113 103.85 46.0409C103.255 46.3885 102.385 46.2682 101.99 46.2449L100.824 47.0806C94.1753 40.0763 85.1235 35.5982 75.3766 34.7283C75.3494 34.3179 75.3137 33.5761 75.3046 33.3529C74.9056 32.9693 74.4235 32.6418 74.3024 31.813C74.1691 30.1569 74.3917 28.3752 74.6496 26.2258C74.7919 25.2216 75.0284 24.3873 75.0688 23.2973C75.078 23.0495 75.0632 22.69 75.0622 22.4221C75.062 20.5322 73.6898 18.9998 71.9968 19ZM68.1585 42.8886L67.248 59.0447L67.1825 59.0776C67.1214 60.5229 65.9375 61.677 64.4839 61.677C63.8884 61.677 63.3388 61.4849 62.8922 61.1571L62.866 61.1703L49.6807 51.7794C53.7331 47.7759 58.9164 44.8172 64.89 43.4546C65.9812 43.2056 67.0719 43.0209 68.1585 42.8886ZM75.8417 42.8886C82.8159 43.7504 89.2657 46.9232 94.2081 51.786L81.108 61.1176L81.0621 61.0979C79.8994 61.9512 78.2611 61.7394 77.3548 60.5978C76.9835 60.1301 76.7887 59.5801 76.7653 59.0249L76.7522 59.0184L75.8417 42.8886ZM44.8991 57.814L56.9382 68.633L56.9251 68.6988C58.0117 69.6479 58.172 71.2949 57.2657 72.4368C56.8944 72.9045 56.3975 73.2182 55.8639 73.3647L55.8508 73.4173L40.4188 77.8923C39.6334 70.6765 41.3261 63.6621 44.8991 57.814ZM99.0094 57.8206C100.798 60.7336 102.153 63.9871 102.959 67.5143C103.756 70.9991 103.956 74.4778 103.627 77.8397L88.1166 73.3515L88.1035 73.2857C86.7145 72.9043 85.8609 71.4848 86.1843 70.0611C86.3168 69.4778 86.6249 68.9844 87.0423 68.6198L87.0358 68.5869L99.0094 57.8206ZM69.5274 69.4688H74.4596L77.5251 73.3186L76.4247 78.1226L71.9968 80.2614L67.5558 78.1161L66.4554 73.312L69.5274 69.4688ZM85.3393 82.6437C85.5489 82.6331 85.7576 82.652 85.9616 82.6898L85.9878 82.6569L101.95 85.3682C99.6142 91.9624 95.144 97.675 89.1711 101.498L82.9747 86.4606L82.9944 86.4343C82.4252 85.1055 82.9948 83.5472 84.3044 82.9135C84.6397 82.7513 84.99 82.6614 85.3393 82.6437ZM58.5298 82.7095C59.748 82.7267 60.8406 83.5761 61.1237 84.822C61.2562 85.4052 61.1917 85.9831 60.973 86.4935L61.0189 86.5527L54.888 101.439C49.1559 97.7432 44.5904 92.2099 42.1481 85.4208L57.9731 82.7227L57.9993 82.7556C58.1763 82.7229 58.3558 82.7071 58.5298 82.7095ZM71.8986 89.2312C72.3229 89.2155 72.7534 89.303 73.1627 89.501C73.6992 89.7606 74.1136 90.1692 74.3745 90.6592H74.4334L82.2346 104.821C81.2221 105.162 80.1813 105.454 79.1167 105.697C73.1505 107.058 67.2032 106.645 61.818 104.802L69.5995 90.6658H69.6126C70.0795 89.7888 70.965 89.2656 71.8986 89.2312Z" fill="white" stroke="white" stroke-width="0.25"/>
<defs>
<linearGradient id="paint0_linear_681_2845" x1="10" y1="15.5" x2="144" y2="131.5" gradientUnits="userSpaceOnUse">
<stop stop-color="#4D87FF"/>
<stop offset="1" stop-color="#0547D0"/>
</linearGradient>
</defs>
</svg>
 - keysOrder: [["apiVersion"], ["appVersion"], ["kind"], ["metadata"], ["metadata", "name"], ["spec", "storageClass"], ["spec", "nodeGroups"], ["spec", "nodeGroups", "md0"], ["spec", "nodeGroups", "md0", "minReplicas"], ["spec", "nodeGroups", "md0", "maxReplicas"], ["spec", "nodeGroups", "md0", "instanceType"], ["spec", "nodeGroups", "md0", "ephemeralStorage"], ["spec", "nodeGroups", "md0", "roles"], ["spec", "nodeGroups", "md0", "resources"], ["spec", "nodeGroups", "md0", "gpus"], ["spec", "version"], ["spec", "host"], ["spec", "addons"], ["spec", "addons", "certManager"], ["spec", "addons", "certManager", "enabled"], ["spec", "addons", "certManager", "valuesOverride"], ["spec", "addons", "cilium"], ["spec", "addons", "cilium", "valuesOverride"], ["spec", "addons", "gatewayAPI"], ["spec", "addons", "gatewayAPI", "enabled"], ["spec", "addons", "ingressNginx"], ["spec", "addons", "ingressNginx", "enabled"], ["spec", "addons", "ingressNginx", "exposeMethod"], ["spec", "addons", "ingressNginx", "hosts"], ["spec", "addons", "ingressNginx", "valuesOverride"], ["spec", "addons", "gpuOperator"], ["spec", "addons", "gpuOperator", "enabled"], ["spec", "addons", "gpuOperator", "valuesOverride"], ["spec", "addons", "fluxcd"], ["spec", "addons", "fluxcd", "enabled"], ["spec", "addons", "fluxcd", "valuesOverride"], ["spec", "addons", "monitoringAgents"], ["spec", "addons", "monitoringAgents", "enabled"], ["spec", "addons", "monitoringAgents", "valuesOverride"], ["spec", "addons", "verticalPodAutoscaler"], ["spec", "addons", "verticalPodAutoscaler", "valuesOverride"], ["spec", "addons", "velero"], ["spec", "addons", "velero", "enabled"], ["spec", "addons", "velero", "valuesOverride"], ["spec", "addons", "coredns"], ["spec", "addons", "coredns", "valuesOverride"], ["spec", "controlPlane"], ["spec", "controlPlane", "replicas"], ["spec", "controlPlane", "apiServer"], ["spec", "controlPlane", "apiServer", "resources"], ["spec", "controlPlane", "apiServer", "resourcesPreset"], ["spec", "controlPlane", "controllerManager"], ["spec", "controlPlane", "controllerManager", "resources"], ["spec", "controlPlane", "controllerManager", "resourcesPreset"], ["spec", "controlPlane", "scheduler"], ["spec", "controlPlane", "scheduler", "resources"], ["spec", "controlPlane", "scheduler", "resourcesPreset"], ["spec", "controlPlane", "konnectivity"], ["spec", "controlPlane", "konnectivity", "server"], ["spec", "controlPlane", "konnectivity", "server", "resources"], ["spec", "controlPlane", "konnectivity", "server", "resourcesPreset"], ["spec", "images"], ["spec", "images", "waitForKubeconfig"]] + keysOrder: [["apiVersion"], ["appVersion"], ["kind"], ["metadata"], ["metadata", "name"], ["spec", "storageClass"], ["spec", "nodeGroups"], ["spec", "nodeGroups", "md0"], ["spec", "nodeGroups", "md0", "minReplicas"], ["spec", "nodeGroups", "md0", "maxReplicas"], ["spec", "nodeGroups", "md0", "instanceType"], ["spec", "nodeGroups", "md0", "ephemeralStorage"], ["spec", "nodeGroups", "md0", "roles"], ["spec", "nodeGroups", "md0", "resources"], ["spec", "nodeGroups", "md0", "gpus"], ["spec", "version"], ["spec", "host"], ["spec", "addons"], ["spec", "addons", "certManager"], ["spec", "addons", "certManager", "enabled"], ["spec", "addons", "certManager", "valuesOverride"], ["spec", "addons", "cilium"], ["spec", "addons", "cilium", "valuesOverride"], ["spec", "addons", "gatewayAPI"], ["spec", "addons", "gatewayAPI", "enabled"], ["spec", "addons", "ingressNginx"], ["spec", "addons", "ingressNginx", "enabled"], ["spec", "addons", "ingressNginx", "exposeMethod"], ["spec", "addons", "ingressNginx", "hosts"], ["spec", "addons", "ingressNginx", "valuesOverride"], ["spec", "addons", "gpuOperator"], ["spec", "addons", "gpuOperator", "enabled"], ["spec", "addons", "gpuOperator", "valuesOverride"], ["spec", "addons", "hami"], ["spec", "addons", "hami", "enabled"], ["spec", "addons", "hami", "valuesOverride"], ["spec", "addons", "fluxcd"], ["spec", "addons", "fluxcd", "enabled"], ["spec", "addons", "fluxcd", "valuesOverride"], ["spec", "addons", "monitoringAgents"], ["spec", "addons", "monitoringAgents", "enabled"], ["spec", "addons", "monitoringAgents", "valuesOverride"], ["spec", "addons", "verticalPodAutoscaler"], ["spec", "addons", "verticalPodAutoscaler", "valuesOverride"], ["spec", "addons", "velero"], ["spec", "addons", "velero", "enabled"], ["spec", "addons", "velero", "valuesOverride"], ["spec", "addons", "coredns"], ["spec", "addons", "coredns", "valuesOverride"], ["spec", "controlPlane"], ["spec", "controlPlane", "replicas"], ["spec", "controlPlane", "apiServer"], ["spec", "controlPlane", "apiServer", "resources"], ["spec", "controlPlane", "apiServer", "resourcesPreset"], ["spec", "controlPlane", "controllerManager"], ["spec", "controlPlane", "controllerManager", "resources"], ["spec", "controlPlane", "controllerManager", "resourcesPreset"], ["spec", "controlPlane", "scheduler"], ["spec", "controlPlane", "scheduler", "resources"], ["spec", "controlPlane", "scheduler", "resourcesPreset"], ["spec", "controlPlane", "konnectivity"], ["spec", "controlPlane", "konnectivity", "server"], ["spec", "controlPlane", "konnectivity", "server", "resources"], ["spec", "controlPlane", "konnectivity", "server", "resourcesPreset"], ["spec", "images"], ["spec", "images", "waitForKubeconfig"]] secrets: exclude: [] include: diff --git a/packages/system/lineage-controller-webhook/Makefile b/packages/system/lineage-controller-webhook/Makefile index d5bada31..cdc1cf01 100644 --- a/packages/system/lineage-controller-webhook/Makefile +++ b/packages/system/lineage-controller-webhook/Makefile @@ -4,8 +4,13 @@ NAMESPACE=cozy-system include ../../../hack/common-envs.mk include ../../../hack/package.mk +.PHONY: test + image: image-lineage-controller-webhook +test: + helm unittest . + image-lineage-controller-webhook: docker buildx build -f images/lineage-controller-webhook/Dockerfile ../../.. \ --tag $(REGISTRY)/lineage-controller-webhook:$(call settag,$(TAG)) \ diff --git a/packages/system/lineage-controller-webhook/README.md b/packages/system/lineage-controller-webhook/README.md new file mode 100644 index 00000000..1e43c9c8 --- /dev/null +++ b/packages/system/lineage-controller-webhook/README.md @@ -0,0 +1,67 @@ +# lineage-controller-webhook + +Cozystack system package for the **lineage controller webhook** — a mutating +admission webhook that stamps "lineage" labels onto tenant workloads, linking +each resource back to the Cozystack `Application` that ultimately owns it. + +The webhook intercepts CREATE and UPDATE on `pods`, `secrets`, `services`, +`persistentvolumeclaims`, `ingresses.networking.k8s.io`, and +`workloadmonitors.cozystack.io` outside system namespaces. For each request it +walks the ownership graph upward (Kubernetes `ownerReferences`, then the +`HelmRelease` Flux installed the resource with, then the Cozystack +`Application`-derived CRD that produced the HelmRelease) and writes the +discovered application's group, kind and name as labels +(`apps.cozystack.io/application.{group,kind,name}`) on the incoming object. +Those labels let the aggregated API server, the Cozystack dashboard, and other +lineage-aware consumers reason about which application a resource belongs to. + +The webhook serves TLS on port 9443 with a cert-manager issued certificate, and +runs with `failurePolicy: Fail` — every CREATE/UPDATE on the resources above +is gated on the webhook being reachable. + +## Topology + +The chart ships a single shape, modelled on `cozystack-api`: + +- **Deployment** with two replicas (override via `replicas`). +- **Soft `nodeAffinity`** preferring `node-role.kubernetes.io/control-plane` + (`Exists`, so it matches both Talos's empty value and k3s/kubeadm's `"true"`). + Soft means: the pod lands on a control-plane node when one is reachable, and + on any worker otherwise — no override needed for managed Kubernetes, + Cozy-in-Cozy tenants, or any other cluster where control-plane nodes aren't + visible. +- **Permissive `tolerations`** (`[{operator: Exists}]`) so the pod can land on + tainted control-plane nodes when the soft affinity is satisfiable. +- **Soft `podAntiAffinity`** on `kubernetes.io/hostname` so replicas spread + across nodes when possible (best-effort). +- **`PodDisruptionBudget`** with `maxUnavailable: 1`, unconditional. At + `replicas: 1` it's a useful no-op (allows full drain); at `replicas >= 2` + it caps disruption to one pod. +- **`Service` with `spec.trafficDistribution: PreferClose`** so the apiserver + prefers a webhook endpoint on its own node when one exists, and transparently + falls over to a remote endpoint otherwise. Requires Kubernetes ≥ 1.31; on + older clusters the field is silently ignored and traffic uses default + cluster-wide distribution. + +## Parameters + +All keys live under the top-level `lineageControllerWebhook:` map. + +| Name | Description | Value | +| ----------------------------- | --------------------------------------------------------- | -------------------------------------------------------------------------------- | +| `image` | Container image (digest-pinned) | `ghcr.io/cozystack/cozystack/lineage-controller-webhook:v1.3.0@sha256:e898…6fb0` | +| `debug` | Enable `--zap-log-level=debug` instead of `info` | `false` | +| `replicas` | Deployment replica count | `2` | +| `localK8sAPIEndpoint.enabled` | **Deprecated.** See note below. | `false` | + +### `localK8sAPIEndpoint.enabled` (deprecated) + +When enabled, this injects `KUBERNETES_SERVICE_HOST=status.hostIP` and +`KUBERNETES_SERVICE_PORT=6443` so the webhook talks to the kube-apiserver on +its own node. It was originally added to avoid latency on the +webhook-to-apiserver path, but it is only valid when the pod is actually +scheduled on an apiserver-bearing node — which the chart's soft control-plane +affinity no longer guarantees. With this flag enabled and the pod scheduled +off a control-plane node, the controller will crash-loop dialing a non- +apiserver hostIP. Slated for removal once the latency motivation is addressed +in the webhook itself; **leave disabled**. diff --git a/packages/system/lineage-controller-webhook/templates/pdb.yaml b/packages/system/lineage-controller-webhook/templates/pdb.yaml new file mode 100644 index 00000000..0786c8ff --- /dev/null +++ b/packages/system/lineage-controller-webhook/templates/pdb.yaml @@ -0,0 +1,9 @@ +apiVersion: policy/v1 +kind: PodDisruptionBudget +metadata: + name: lineage-controller-webhook +spec: + maxUnavailable: 1 + selector: + matchLabels: + app: lineage-controller-webhook diff --git a/packages/system/lineage-controller-webhook/templates/service.yaml b/packages/system/lineage-controller-webhook/templates/service.yaml index c5df90fb..fbb3ad54 100644 --- a/packages/system/lineage-controller-webhook/templates/service.yaml +++ b/packages/system/lineage-controller-webhook/templates/service.yaml @@ -5,7 +5,7 @@ metadata: labels: app: lineage-controller-webhook spec: - internalTrafficPolicy: Local + trafficDistribution: PreferClose type: ClusterIP ports: - port: 443 diff --git a/packages/system/lineage-controller-webhook/templates/daemonset.yaml b/packages/system/lineage-controller-webhook/templates/workload.yaml similarity index 67% rename from packages/system/lineage-controller-webhook/templates/daemonset.yaml rename to packages/system/lineage-controller-webhook/templates/workload.yaml index 860aee6e..097919c6 100644 --- a/packages/system/lineage-controller-webhook/templates/daemonset.yaml +++ b/packages/system/lineage-controller-webhook/templates/workload.yaml @@ -1,10 +1,11 @@ apiVersion: apps/v1 -kind: DaemonSet +kind: Deployment metadata: name: lineage-controller-webhook labels: app: lineage-controller-webhook spec: + replicas: {{ .Values.lineageControllerWebhook.replicas }} selector: matchLabels: app: lineage-controller-webhook @@ -13,13 +14,25 @@ spec: labels: app: lineage-controller-webhook spec: - {{- with .Values.lineageControllerWebhook.nodeSelector }} - nodeSelector: - {{- toYaml . | nindent 8 }} - {{- end }} + serviceAccountName: lineage-controller-webhook tolerations: - operator: Exists - serviceAccountName: lineage-controller-webhook + affinity: + nodeAffinity: + preferredDuringSchedulingIgnoredDuringExecution: + - weight: 100 + preference: + matchExpressions: + - key: node-role.kubernetes.io/control-plane + operator: Exists + podAntiAffinity: + preferredDuringSchedulingIgnoredDuringExecution: + - weight: 100 + podAffinityTerm: + topologyKey: kubernetes.io/hostname + labelSelector: + matchLabels: + app: lineage-controller-webhook containers: - name: lineage-controller-webhook image: "{{ .Values.lineageControllerWebhook.image }}" diff --git a/packages/system/lineage-controller-webhook/tests/service_test.yaml b/packages/system/lineage-controller-webhook/tests/service_test.yaml new file mode 100644 index 00000000..3cdbddd8 --- /dev/null +++ b/packages/system/lineage-controller-webhook/tests/service_test.yaml @@ -0,0 +1,30 @@ +suite: lineage-controller-webhook service +templates: + - templates/service.yaml + +release: + name: lineage-controller-webhook + namespace: cozy-system + +tests: + - it: renders a ClusterIP service on 443 -> 9443 with PreferClose traffic distribution + asserts: + - isKind: + of: Service + - equal: + path: spec.type + value: ClusterIP + - equal: + path: spec.trafficDistribution + value: PreferClose + - notExists: + path: spec.internalTrafficPolicy + - equal: + path: spec.ports[0].port + value: 443 + - equal: + path: spec.ports[0].targetPort + value: 9443 + - equal: + path: spec.selector.app + value: lineage-controller-webhook diff --git a/packages/system/lineage-controller-webhook/tests/workload_test.yaml b/packages/system/lineage-controller-webhook/tests/workload_test.yaml new file mode 100644 index 00000000..bcb24014 --- /dev/null +++ b/packages/system/lineage-controller-webhook/tests/workload_test.yaml @@ -0,0 +1,99 @@ +suite: lineage-controller-webhook workload + PDB +templates: + - templates/workload.yaml + - templates/pdb.yaml + +release: + name: lineage-controller-webhook + namespace: cozy-system + +tests: + # ---- Default Deployment shape ---------------------------------------------- + + - it: defaults render a 2-replica Deployment with soft control-plane nodeAffinity, soft podAntiAffinity, and Exists tolerations + template: templates/workload.yaml + asserts: + - isKind: + of: Deployment + - equal: + path: spec.replicas + value: 2 + - equal: + path: spec.template.spec.tolerations + value: + - operator: Exists + - equal: + path: spec.template.spec.affinity.nodeAffinity.preferredDuringSchedulingIgnoredDuringExecution + value: + - weight: 100 + preference: + matchExpressions: + - key: node-role.kubernetes.io/control-plane + operator: Exists + - notExists: + path: spec.template.spec.affinity.nodeAffinity.requiredDuringSchedulingIgnoredDuringExecution + - equal: + path: spec.template.spec.affinity.podAntiAffinity.preferredDuringSchedulingIgnoredDuringExecution[0].podAffinityTerm.topologyKey + value: kubernetes.io/hostname + + - it: defaults inject no localK8sAPIEndpoint env vars (knob is opt-in) + template: templates/workload.yaml + asserts: + - notExists: + path: spec.template.spec.containers[0].env + + # ---- PDB -------------------------------------------------------------------- + + - it: PDB renders unconditionally with maxUnavailable=1 at the default replica count + template: templates/pdb.yaml + asserts: + - isKind: + of: PodDisruptionBudget + - equal: + path: spec.maxUnavailable + value: 1 + - notExists: + path: spec.minAvailable + + - it: PDB still renders at replicas=1 (no-op, but consistent shape) + set: + lineageControllerWebhook.replicas: 1 + template: templates/pdb.yaml + asserts: + - isKind: + of: PodDisruptionBudget + - equal: + path: spec.maxUnavailable + value: 1 + + # ---- Replica override ------------------------------------------------------- + + - it: replicas value drives spec.replicas + set: + lineageControllerWebhook.replicas: 5 + template: templates/workload.yaml + asserts: + - equal: + path: spec.replicas + value: 5 + + # ---- Deprecated localK8sAPIEndpoint knob ----------------------------------- + + - it: localK8sAPIEndpoint.enabled=true injects KUBERNETES_SERVICE_HOST and PORT env vars + set: + lineageControllerWebhook.localK8sAPIEndpoint.enabled: true + template: templates/workload.yaml + asserts: + - contains: + path: spec.template.spec.containers[0].env + content: + name: KUBERNETES_SERVICE_HOST + valueFrom: + fieldRef: + apiVersion: v1 + fieldPath: status.hostIP + - contains: + path: spec.template.spec.containers[0].env + content: + name: KUBERNETES_SERVICE_PORT + value: "6443" diff --git a/packages/system/lineage-controller-webhook/values.yaml b/packages/system/lineage-controller-webhook/values.yaml index c5671548..e56d332d 100644 --- a/packages/system/lineage-controller-webhook/values.yaml +++ b/packages/system/lineage-controller-webhook/values.yaml @@ -1,10 +1,13 @@ lineageControllerWebhook: image: ghcr.io/cozystack/cozystack/lineage-controller-webhook:v1.3.0@sha256:e8984709686a5eaf19b89da378d7b8c688ea5607e0783a88d9c9e4ccfca96fb0 debug: false + replicas: 2 + # DEPRECATED. Injects KUBERNETES_SERVICE_HOST=status.hostIP and + # KUBERNETES_SERVICE_PORT=6443 so the webhook talks to the apiserver on its + # own node — only valid when the pod is actually scheduled on an apiserver- + # bearing node. Now that the chart's nodeAffinity to control-plane is soft + # (preferred, not required), the pod can land off-control-plane and crash- + # loop dialing a non-apiserver hostIP. Slated for removal once webhook + # performance work obviates the locality motivation. Leave disabled. localK8sAPIEndpoint: - enabled: true - # nodeSelector for the DaemonSet - # Talos uses empty value: "node-role.kubernetes.io/control-plane": "" - # Generic k8s (k3s, kubeadm) uses: "node-role.kubernetes.io/control-plane": "true" - nodeSelector: - node-role.kubernetes.io/control-plane: "" + enabled: false diff --git a/packages/system/linstor-gui/Makefile b/packages/system/linstor-gui/Makefile index e202f4cc..b408491d 100644 --- a/packages/system/linstor-gui/Makefile +++ b/packages/system/linstor-gui/Makefile @@ -12,18 +12,13 @@ image: image-linstor-gui image-linstor-gui: docker buildx build images/linstor-gui \ - --provenance false \ - --builder=$(BUILDER) \ - --platform=linux/amd64,linux/arm64 \ --build-arg LINSTOR_GUI_VERSION=$(LINSTOR_GUI_VERSION) \ --tag $(REGISTRY)/linstor-gui:$(call settag,$(LINSTOR_GUI_VERSION)) \ --tag $(REGISTRY)/linstor-gui:$(call settag,$(LINSTOR_GUI_VERSION)-$(TAG)) \ --cache-from type=registry,ref=$(REGISTRY)/linstor-gui:latest \ --cache-to type=inline \ --metadata-file images/linstor-gui.json \ - --push=$(PUSH) \ - --label "org.opencontainers.image.source=https://github.com/cozystack/cozystack" \ - --load=$(LOAD) + $(BUILDX_ARGS) REPOSITORY="$(REGISTRY)/linstor-gui" \ yq -i '.image.repository = strenv(REPOSITORY)' values.yaml TAG="$(call settag,$(LINSTOR_GUI_VERSION))@$$(yq e '."containerimage.digest"' images/linstor-gui.json -o json -r)" \ diff --git a/packages/system/linstor/Makefile b/packages/system/linstor/Makefile index 2ad4def9..8d566678 100644 --- a/packages/system/linstor/Makefile +++ b/packages/system/linstor/Makefile @@ -5,7 +5,7 @@ include ../../../hack/common-envs.mk include ../../../hack/package.mk LINSTOR_VERSION ?= 1.33.2 -LINSTOR_CSI_VERSION ?= v1.10.5 +LINSTOR_CSI_VERSION ?= v1.10.6 image: image-piraeus-server image-linstor-csi diff --git a/packages/system/linstor/images/linstor-csi/Dockerfile b/packages/system/linstor/images/linstor-csi/Dockerfile index 6cfd44aa..df5d179d 100644 --- a/packages/system/linstor/images/linstor-csi/Dockerfile +++ b/packages/system/linstor/images/linstor-csi/Dockerfile @@ -1,6 +1,6 @@ FROM golang:1.25 AS builder -ARG VERSION=v1.10.5 +ARG VERSION=v1.10.6 ARG LINSTOR_WAIT_UNTIL_VERSION=v0.3.1 ARG TARGETARCH ARG TARGETOS diff --git a/packages/system/linstor/images/linstor-csi/patches/001-relocate-after-clone-restore.diff b/packages/system/linstor/images/linstor-csi/patches/001-relocate-after-clone-restore.diff index ff0aec0c..0e272670 100644 --- a/packages/system/linstor/images/linstor-csi/patches/001-relocate-after-clone-restore.diff +++ b/packages/system/linstor/images/linstor-csi/patches/001-relocate-after-clone-restore.diff @@ -1,25 +1,7 @@ diff --git a/pkg/client/linstor.go b/pkg/client/linstor.go -index f544493..98e7fde 100644 +index ac4651d..46d565d 100644 --- a/pkg/client/linstor.go +++ b/pkg/client/linstor.go -@@ -181,7 +181,7 @@ func LogLevel(s string) func(*Linstor) error { - func (s *Linstor) ListAllWithStatus(ctx context.Context) ([]volume.VolumeStatus, error) { - var vols []volume.VolumeStatus - -- resourcesByName := make(map[string][]lapi.Resource) -+ resourcesByName := make(map[string][]lapi.ResourceWithVolumes) - - resDefs, err := s.client.ResourceDefinitions.GetAll(ctx, lapi.RDGetAllRequest{WithVolumeDefinitions: true}) - if err != nil { -@@ -194,7 +194,7 @@ func (s *Linstor) ListAllWithStatus(ctx context.Context) ([]volume.VolumeStatus, - } - - for i := range allResources { -- resourcesByName[allResources[i].Name] = append(resourcesByName[allResources[i].Name], allResources[i].Resource) -+ resourcesByName[allResources[i].Name] = append(resourcesByName[allResources[i].Name], allResources[i]) - } - - for _, rd := range resDefs { @@ -462,6 +462,14 @@ func (s *Linstor) Clone(ctx context.Context, vol, src *volume.Info, params *volu return err } @@ -35,7 +17,7 @@ index f544493..98e7fde 100644 logger.Debug("reconcile extra properties") err = s.client.ResourceDefinitions.Modify(ctx, vol.ID, lapi.GenericPropsModify{OverrideProps: vol.Properties}) -@@ -1280,6 +1288,14 @@ func (s *Linstor) VolFromSnap(ctx context.Context, snap *volume.Snapshot, vol *v +@@ -1297,6 +1305,14 @@ func (s *Linstor) VolFromSnap(ctx context.Context, snap *volume.Snapshot, vol *v return err } @@ -50,7 +32,7 @@ index f544493..98e7fde 100644 logger.Debug("reconcile extra properties") err = s.client.ResourceDefinitions.Modify(ctx, vol.ID, lapi.GenericPropsModify{OverrideProps: vol.Properties}) -@@ -1470,9 +1486,8 @@ func (s *Linstor) reconcileSnapshotResources(ctx context.Context, snapshot *volu +@@ -1487,9 +1503,8 @@ func (s *Linstor) reconcileSnapshotResources(ctx context.Context, snapshot *volu return fmt.Errorf("snapshot '%s' not deployed on any node", snap.Name) } @@ -62,7 +44,7 @@ index f544493..98e7fde 100644 for _, snapNode := range snap.Nodes { if err := s.NodeAvailable(ctx, snapNode); err != nil { logger.WithField("selected node candidate", snapNode).WithError(err).Debug("node is not available") -@@ -1480,13 +1495,23 @@ func (s *Linstor) reconcileSnapshotResources(ctx context.Context, snapshot *volu +@@ -1497,13 +1512,23 @@ func (s *Linstor) reconcileSnapshotResources(ctx context.Context, snapshot *volu } if slices.Contains(preferredNodes, snapNode) { @@ -92,7 +74,7 @@ index f544493..98e7fde 100644 } if selectedNode == "" { -@@ -1679,6 +1704,114 @@ func (s *Linstor) reconcileResourcePlacement(ctx context.Context, vol *volume.In +@@ -1696,6 +1721,114 @@ func (s *Linstor) reconcileResourcePlacement(ctx context.Context, vol *volume.In return nil } @@ -207,40 +189,8 @@ index f544493..98e7fde 100644 // FindSnapsByID searches the snapshot in the backend func (s *Linstor) FindSnapsByID(ctx context.Context, id string) ([]*volume.Snapshot, error) { snapshotId, err := volume.ParseSnapshotId(id) -@@ -2173,7 +2306,7 @@ func (s *Linstor) Status(ctx context.Context, volId string) ([]string, *csi.Volu - "volume": volId, - }).Debug("getting assignments") - -- ress, err := s.client.Resources.GetAll(ctx, volId) -+ ress, err := s.client.Resources.GetResourceView(ctx, &lapi.ListOpts{Resource: []string{volId}}) - if err != nil { - return nil, nil, fmt.Errorf("failed to list resources for '%s': %w", volId, err) - } -@@ -2261,13 +2394,20 @@ func GetSnapshotRemoteAndReadiness(snap *lapi.Snapshot) (string, bool, error) { - return "", slices.Contains(snap.Flags, lapiconsts.FlagSuccessful), nil - } - --func NodesAndConditionFromResources(ress []lapi.Resource) ([]string, *csi.VolumeCondition) { -+func NodesAndConditionFromResources(ress []lapi.ResourceWithVolumes) ([]string, *csi.VolumeCondition) { - var allNodes, abnormalNodes []string - - for i := range ress { - res := &ress[i] - -- allNodes = append(allNodes, res.NodeName) -+ // A resource is a CSI publish target if any of its volumes were created -+ // by ControllerPublishVolume, identified by the temporary-diskless-attach property. -+ if slices.ContainsFunc(res.Volumes, func(v lapi.Volume) bool { -+ createdFor, ok := v.Props[linstor.PropertyCreatedFor] -+ return ok && createdFor == linstor.CreatedForTemporaryDisklessAttach -+ }) { -+ allNodes = append(allNodes, res.NodeName) -+ } - - if res.State == nil { - abnormalNodes = append(abnormalNodes, res.NodeName) diff --git a/pkg/volume/parameter.go b/pkg/volume/parameter.go -index 39acd95..aed18ab 100644 +index 39acd95..54d1dfb 100644 --- a/pkg/volume/parameter.go +++ b/pkg/volume/parameter.go @@ -50,6 +50,7 @@ const ( diff --git a/packages/system/linstor/images/linstor-csi/patches/002-protocol-c-override-for-dual-attach.diff b/packages/system/linstor/images/linstor-csi/patches/002-protocol-c-override-for-dual-attach.diff new file mode 100644 index 00000000..279c636b --- /dev/null +++ b/packages/system/linstor/images/linstor-csi/patches/002-protocol-c-override-for-dual-attach.diff @@ -0,0 +1,132 @@ +diff --git a/pkg/client/linstor.go b/pkg/client/linstor.go +index ac4651d..9d75c79 100644 +--- a/pkg/client/linstor.go ++++ b/pkg/client/linstor.go +@@ -653,11 +653,35 @@ func (s *Linstor) Attach(ctx context.Context, volId, node string, rwxBlock bool) + } + + if otherResInUse > 0 && rwxBlock { +- rdPropsModify := lapi.GenericPropsModify{OverrideProps: map[string]string{ ++ rdProps := map[string]string{ + linstor.PropertyAllowTwoPrimaries: "yes", +- }} ++ } + +- err = s.client.ResourceDefinitions.Modify(ctx, volId, rdPropsModify) ++ // DRBD requires Protocol C whenever allow-two-primaries is enabled. ++ // If the resource is configured with Protocol A or B (typically ++ // through its resource-group), drbdadm adjust on the satellites ++ // fails with "Protocol C required" (errno 139) and live migration ++ // cannot proceed. Override Protocol to C on the resource-definition ++ // itself so it applies to every connection (including diskless ++ // peers, e.g. a TieBreaker). The marker lets Detach revert exactly ++ // the override we installed without disturbing operator-set props. ++ proto, protoErr := s.getEffectiveDrbdProtocol(ctx, volId) ++ if protoErr != nil { ++ s.log.WithError(protoErr).WithField("volume", volId). ++ Warn("failed to determine effective DRBD protocol; skipping Protocol=C override for dual-attach") ++ } else if proto == "A" || proto == "B" { ++ s.log.WithFields(logrus.Fields{ ++ "volume": volId, ++ "protocol": proto, ++ }).Info("installing Protocol=C override on resource-definition for dual-attach") ++ ++ rdProps[linstor.PropertyDrbdNetProtocol] = "C" ++ rdProps[linstor.PropertyCsiProtocolOverride] = "yes" ++ } ++ ++ err = s.client.ResourceDefinitions.Modify(ctx, volId, lapi.GenericPropsModify{ ++ OverrideProps: rdProps, ++ }) + if err != nil { + return "", err + } +@@ -774,11 +798,26 @@ func (s *Linstor) Detach(ctx context.Context, volId, node string) error { + } + + if resInUse == 1 { +- rdPropsModify := lapi.GenericPropsModify{DeleteProps: []string{ +- linstor.PropertyAllowTwoPrimaries, +- }} ++ deleteProps := []string{linstor.PropertyAllowTwoPrimaries} ++ ++ // Drop the Protocol=C override only if Attach installed it (gated by ++ // our marker), so an operator-set Protocol property on the ++ // resource-definition is preserved. ++ rd, getErr := s.client.ResourceDefinitions.Get(ctx, volId) ++ if getErr != nil { ++ log.WithError(getErr).Warn("failed to fetch resource-definition props; skipping Protocol=C override removal") ++ } else if rd.Props[linstor.PropertyCsiProtocolOverride] != "" { ++ log.Info("removing Protocol=C override from resource-definition") + +- err = s.client.ResourceDefinitions.Modify(ctx, volId, rdPropsModify) ++ deleteProps = append(deleteProps, ++ linstor.PropertyDrbdNetProtocol, ++ linstor.PropertyCsiProtocolOverride, ++ ) ++ } ++ ++ err = s.client.ResourceDefinitions.Modify(ctx, volId, lapi.GenericPropsModify{ ++ DeleteProps: deleteProps, ++ }) + if err != nil { + return nil404(err) + } +@@ -805,6 +844,33 @@ func (s *Linstor) Detach(ctx context.Context, volId, node string) error { + return nil404(s.client.Resources.Delete(ctx, volId, node)) + } + ++// getEffectiveDrbdProtocol returns the DRBD network protocol ("A", "B" or "C") ++// that LINSTOR will hand to drbdadm for the given resource. Resource-definition ++// properties take precedence; otherwise the value is inherited from the ++// resource-group. An empty string is returned when neither level sets the ++// property (LINSTOR's compiled-in default is "C"). ++func (s *Linstor) getEffectiveDrbdProtocol(ctx context.Context, volId string) (string, error) { ++ rd, err := s.client.ResourceDefinitions.Get(ctx, volId) ++ if err != nil { ++ return "", err ++ } ++ ++ if v, ok := rd.Props[linstor.PropertyDrbdNetProtocol]; ok { ++ return v, nil ++ } ++ ++ if rd.ResourceGroupName == "" { ++ return "", nil ++ } ++ ++ rg, err := s.client.ResourceGroups.Get(ctx, rd.ResourceGroupName) ++ if err != nil { ++ return "", err ++ } ++ ++ return rg.Props[linstor.PropertyDrbdNetProtocol], nil ++} ++ + // CapacityBytes returns the amount of free space in the storage pool specified by the params and topology. + func (s *Linstor) CapacityBytes(ctx context.Context, storagePools []string, overProvision *float64, segments map[string]string) (int64, error) { + log := s.log.WithField("storage-pools", storagePools).WithField("segments", segments) +diff --git a/pkg/linstor/const.go b/pkg/linstor/const.go +index 9a5f79c..c8bc9c3 100644 +--- a/pkg/linstor/const.go ++++ b/pkg/linstor/const.go +@@ -44,6 +44,19 @@ const ( + // PropertyAllowTwoPrimaries is DRBD option to allow second primary. Mainly used for live-migration. + PropertyAllowTwoPrimaries = lc.NamespcDrbdNetOptions + "/allow-two-primaries" + ++ // PropertyDrbdNetProtocol is the DRBD network replication protocol option ++ // (A = async, B = semi-sync, C = sync). DRBD requires Protocol C whenever ++ // allow-two-primaries is enabled, otherwise drbdadm adjust fails with ++ // "Protocol C required" (errno 139). ++ PropertyDrbdNetProtocol = lc.NamespcDrbdNetOptions + "/protocol" ++ ++ // PropertyCsiProtocolOverride marks a resource-connection where this driver ++ // has installed a temporary Protocol=C override to make a Protocol-A/B ++ // resource compatible with allow-two-primaries during live migration. The ++ // marker lets us safely remove only the overrides we set, without touching ++ // connection-level overrides installed by the operator. ++ PropertyCsiProtocolOverride = lc.NamespcAuxiliary + "/csi-protocol-override" ++ + // CreatedForTemporaryDisklessAttach marks a resource as temporary, i.e. it should be removed after it is no longer + // needed. + CreatedForTemporaryDisklessAttach = "temporary-diskless-attach" diff --git a/packages/system/monitoring-agents/alerts/gpu-recording.rules.yaml b/packages/system/monitoring-agents/alerts/gpu-recording.rules.yaml new file mode 100644 index 00000000..5e0f9ea0 --- /dev/null +++ b/packages/system/monitoring-agents/alerts/gpu-recording.rules.yaml @@ -0,0 +1,172 @@ +apiVersion: operator.victoriametrics.com/v1beta1 +kind: VMRule +metadata: + name: alerts-gpu-recording.rules +spec: + groups: + - name: gpu.recording.cluster.1m + interval: 1m + params: {} + rules: + - record: cluster:gpu_count:total + expr: count(group by (UUID) (DCGM_FI_DEV_GPU_UTIL)) + # Kube-allocated GPU count: GPUs requested by *active* pods — Pending + # and Running only. Source of truth for "what tenants asked for" — used + # for capacity planning and billing. Includes system pods so it stays + # consistent with cluster:gpu_count:total when computing :free. + # + # Phase filter via group_left() on kube_pod_status_phase matches the + # canonical pattern from k8s.rules.container_resource.yaml. Without it + # kube-state-metrics keeps series for Failed/Succeeded pods until the + # apiserver garbage-collects them, which inflates billing hours and + # can push :free negative. + - record: cluster:gpu_count:allocated + expr: | + sum( + kube_pod_container_resource_requests{resource="nvidia_com_gpu"} + * on (namespace, pod) group_left() max by (namespace, pod) ( + kube_pod_status_phase{phase=~"Pending|Running"} == 1 + ) + ) + # clamp_min guards against transients at DCGM/kube-state-metrics + # restart where cluster:gpu_count:total dips before :allocated catches + # up, or rare label drift between the two sources. + - record: cluster:gpu_count:free + expr: clamp_min(cluster:gpu_count:total - (cluster:gpu_count:allocated or vector(0)), 0) + # Cluster-layer filter asymmetry — intentional, not a bug: + # + # * cluster:gpu_count:total and cluster:gpu_power_watts:sum do NOT + # filter cozy-*/kube-* namespaces because they describe the + # physical hardware fleet. A GPU draws power and occupies a + # slot regardless of which namespace's pod happens to hold it; + # excluding infra pods would under-report raw capacity and + # break capacity-planning math. + # + # No namespace filter — DCGM exporter metrics carry the exporter's + # own namespace, not the workload namespace. Filtering by namespace + # here would silently drop all series. + - record: cluster:gpu_util:avg + expr: avg(DCGM_FI_DEV_GPU_UTIL) + - record: cluster:gpu_power_watts:sum + expr: sum(DCGM_FI_DEV_POWER_USAGE) + + - name: gpu.recording.node.1m + interval: 1m + params: {} + rules: + # Kube-requested GPU count per namespace — billable/allocation view. + # This is the only namespace-level rule that works correctly because + # kube_pod_container_resource_requests carries the real workload + # namespace, unlike DCGM exporter metrics which carry the exporter's + # own namespace (cozy-gpu-operator). + - record: namespace:gpu_count:allocated + expr: | + sum by (namespace) ( + kube_pod_container_resource_requests{resource="nvidia_com_gpu", namespace!="", namespace!~"cozy-.*|kube-.*"} + * on (namespace, pod) group_left() max by (namespace, pod) ( + kube_pod_status_phase{phase=~"Pending|Running"} == 1 + ) + ) + # Hardware metrics aggregated per node (Hostname). DCGM exporter + # metrics carry a capital-H "Hostname" label identifying the node + # but no usable workload namespace — so hardware telemetry is + # aggregated at node/GPU granularity, not namespace. + - record: node:gpu_util:avg + expr: avg by (Hostname) (DCGM_FI_DEV_GPU_UTIL) + - record: node:tensor_active:avg + expr: avg by (Hostname) (DCGM_FI_PROF_PIPE_TENSOR_ACTIVE) + - record: node:fb_used_bytes:sum + expr: sum by (Hostname) (DCGM_FI_DEV_FB_USED) * 1048576 + - record: node:power_watts:sum + expr: sum by (Hostname) (DCGM_FI_DEV_POWER_USAGE) + + # Known gap: there is no lower-bound sanity alert for under-reporting. + # If DCGM ever switches POWER/THERMAL_VIOLATION counter units the other + # direction — e.g. from the current nanoseconds back to microseconds, + # making rate()/1e9 produce values around 0.001 instead of real + # fractions — the throttle signal would silently collapse to near-zero + # instead of plateauing at 1.0. GPUThrottleFractionOverOne below catches + # the >1 drift; the <<1 drift would require correlation with independent + # signals (power draw vs TDP, clock dips) that we do not yet wire up. + # Documented here so future maintainers know this is a known gap, not + # an oversight. + - name: gpu.recording.efficiency.1m + interval: 1m + params: {} + rules: + # Tensor hardware saturation per GPU — the honest "am I using the GPU" + # metric for AI/LLM workloads. Aggregated at GPU level (Hostname + + # gpu + UUID) because DCGM metrics don't carry workload namespace. + # Stored as a 0..1 ratio; consumers multiply by 100 at display time. + - record: gpu:tensor_saturation:avg5m + expr: | + avg by (Hostname, gpu, UUID) ( + avg_over_time(DCGM_FI_PROF_PIPE_TENSOR_ACTIVE[5m]) + ) + + # Power efficiency per GPU — utilization per watt, reveals + # unoptimized workloads. Explicit on(...) pins the matching set + # to GPU-identifying labels. + - record: gpu:util_per_watt:avg5m + expr: | + max by (Hostname, gpu, UUID) ( + avg_over_time(DCGM_FI_DEV_GPU_UTIL[5m]) + ) + / on (Hostname, gpu, UUID) + clamp_min( + max by (Hostname, gpu, UUID) ( + avg_over_time(DCGM_FI_DEV_POWER_USAGE[5m]) + ), + 1 + ) + + # Fraction of time power-throttled (TDP cap) — 1.0 = fully throttled. + # DCGM_FI_DEV_*_VIOLATION is documented as µs but on A10/DCGM 3.x the + # counter grows in nanoseconds in practice — divide by 1e9 to get a + # 0..1 fraction (verified empirically when /1e6 yielded >100× reality). + # clamp_max protects against rate() artefacts at counter resets. + # + # max by (Hostname, gpu, UUID) collapses duplicate series from + # dcgm-exporter's pod-mapping labels (one physical GPU may appear + # under multiple pod/container labels during restarts or under + # MIG/MPS). Throttling is a physical-GPU property; taking max + # keeps consumer avg(...) honest — a naive avg() would dilute the + # signal by the number of pods sharing the GPU. + - record: gpu:power_throttle_fraction:rate5m + expr: clamp_max(max by (Hostname, gpu, UUID) (rate(DCGM_FI_DEV_POWER_VIOLATION[5m]) / 1e9), 1) + + # Fraction of time thermal-throttled. + - record: gpu:thermal_throttle_fraction:rate5m + expr: clamp_max(max by (Hostname, gpu, UUID) (rate(DCGM_FI_DEV_THERMAL_VIOLATION[5m]) / 1e9), 1) + + # Regression watch — the /1e9 divisor on *_VIOLATION was derived + # empirically against DCGM 3.x on A10. If a future DCGM version + # restores the documented µs units, pre-clamp values would exceed 1.0 + # while clamp_max(..., 1) silently masks the drift — recorded + # throttle fractions would plateau at 1.0 and consumers would think + # every GPU is always throttled. This alert fires on that condition + # so we can rescale to /1e6 before dashboards start lying. + - name: gpu.recording.throttle.validation.5m + interval: 5m + params: {} + rules: + - alert: GPUThrottleFractionOverOne + expr: | + max by (Hostname, gpu, UUID) (rate(DCGM_FI_DEV_POWER_VIOLATION[5m]) / 1e9) > 1 + or + max by (Hostname, gpu, UUID) (rate(DCGM_FI_DEV_THERMAL_VIOLATION[5m]) / 1e9) > 1 + for: 15m + labels: + severity: warning + component: gpu-monitoring + annotations: + summary: "DCGM throttle fraction exceeds 1.0 pre-clamp on {{ $labels.Hostname }}/{{ $labels.gpu }}" + description: | + gpu:{power,thermal}_throttle_fraction:rate5m clamps to 1.0, but the raw + rate(DCGM_FI_DEV_*_VIOLATION)/1e9 value is already >1 on this GPU — the + empirical nanosecond assumption is broken. Likely the DCGM exporter + version changed and the counter now grows in its documented µs units. + Re-verify on the exporter Pod and adjust the divisor (/1e9 → /1e6) in + gpu.recording.efficiency.1m before dashboards plateau at 100% throttle. + verified_on: "NVIDIA A10, DCGM 3.x — other GPU families (H100, L40S) may require a different divisor" + runbook: "If this alert fires, DCGM may have changed POWER_VIOLATION/THERMAL_VIOLATION counter units. Recalibrate by comparing raw counter rate against known throttle events; see gpu-recording.rules.yaml comments." diff --git a/packages/system/monitoring/dashboards-infra.list b/packages/system/monitoring/dashboards-infra.list index b5b9f52a..4c7e494d 100644 --- a/packages/system/monitoring/dashboards-infra.list +++ b/packages/system/monitoring/dashboards-infra.list @@ -32,3 +32,8 @@ hubble/overview hubble/dns-namespace hubble/l7-http-metrics hubble/network-overview +gpu/gpu-performance +gpu/gpu-efficiency +gpu/gpu-quotas +gpu/gpu-fleet +gpu/gpu-tenants diff --git a/packages/system/postgres-operator/Makefile b/packages/system/postgres-operator/Makefile index f6d7ddaa..5279f8a7 100644 --- a/packages/system/postgres-operator/Makefile +++ b/packages/system/postgres-operator/Makefile @@ -3,6 +3,9 @@ export NAMESPACE=cozy-$(NAME) include ../../../hack/package.mk +test: + helm unittest . + update: rm -rf charts helm repo add cnpg https://cloudnative-pg.github.io/charts diff --git a/packages/system/postgres-operator/templates/webhook-ready-hook.yaml b/packages/system/postgres-operator/templates/webhook-ready-hook.yaml new file mode 100644 index 00000000..51a8ba81 --- /dev/null +++ b/packages/system/postgres-operator/templates/webhook-ready-hook.yaml @@ -0,0 +1,148 @@ +{{- /* + Post-install gate: block the HelmRelease from reporting Ready until the + cnpg admission webhook actually serves through the cluster Service. Helm + --wait on the controller pod passes once its readinessProbe passes, but + EndpointSlice propagation and kube-proxy/cilium data-plane programming + can lag by a second or two — long enough for any HelmRelease that + depends on postgres-operator (e.g. cozy-keycloak, tenant Postgres apps) + to fire its own install and have kube-apiserver hit the mcluster.cnpg.io + mutating webhook with "dial tcp :443: connect: connection refused". + + The Job uses the apiserver service proxy, which exercises the same + endpoint-resolution and apiserver-initiated pod dial that the admission + webhook path uses. Once /readyz answers through the proxy the data-plane + race is resolved. It does not verify the webhook's TLS CA bundle, so + this gate is scoped to reachability regressions, not cert rotation. + + The service name and port name are hardcoded literals. Upstream cnpg + pins the service name in charts/cloudnative-pg/values.yaml with a + comment "DO NOT CHANGE THE SERVICE NAME as it is currently used to + generate the certificate and can not be configured". The port name is + fixed in charts/cloudnative-pg/templates/service.yaml (ports[0].name: + webhook-server). If a future `make update` ever changes either literal + upstream, the sync-check helm-unittest test + (tests/webhook-ready-hook_test.yaml) renders the subchart Service and + fails if the literal drifts — forcing this template to be updated in + the same change. +*/}} +{{- $_ := required "webhookReady.image.repository must be set to the container image providing kubectl for the post-install readiness Job" .Values.webhookReady.image.repository -}} +{{- $_ := required "webhookReady.image.tag must be set for the post-install readiness Job" .Values.webhookReady.image.tag -}} +{{- /* $svcName and $portName are hardcoded literals; see header comment. */ -}} +{{- $svcName := "cnpg-webhook-service" -}} +{{- /* $portName is the service port NAME, not number — matches ports[0].name in the vendored subchart's Service. */ -}} +{{- $portName := "webhook-server" -}} +{{- $resourceName := printf "https:%s:%s" $svcName $portName -}} +{{- $maxAttempts := .Values.webhookReady.maxAttempts | default 60 -}} +{{- $sleepSeconds := .Values.webhookReady.sleepSeconds | default 2 -}} +{{- /* Derive activeDeadlineSeconds from retries + 60s slack so a values override that raises maxAttempts doesn't get silently cut. */ -}} +{{- $deadline := add (mul (int $maxAttempts) (int $sleepSeconds)) 60 -}} +{{- $image := printf "%s:%s" .Values.webhookReady.image.repository .Values.webhookReady.image.tag -}} +{{- if .Values.webhookReady.image.digest }} +{{- $image = printf "%s@%s" $image .Values.webhookReady.image.digest -}} +{{- end }} +--- +apiVersion: v1 +kind: ServiceAccount +metadata: + name: {{ .Release.Name }}-webhook-ready + namespace: {{ .Release.Namespace }} + annotations: + helm.sh/hook: post-install,post-upgrade + helm.sh/hook-weight: "0" + helm.sh/hook-delete-policy: before-hook-creation,hook-succeeded +--- +apiVersion: rbac.authorization.k8s.io/v1 +kind: Role +metadata: + name: {{ .Release.Name }}-webhook-ready + namespace: {{ .Release.Namespace }} + annotations: + helm.sh/hook: post-install,post-upgrade + helm.sh/hook-weight: "0" + helm.sh/hook-delete-policy: before-hook-creation,hook-succeeded +rules: + - apiGroups: [""] + resources: ["services/proxy"] + resourceNames: [{{ $resourceName | quote }}] + verbs: ["get"] +--- +apiVersion: rbac.authorization.k8s.io/v1 +kind: RoleBinding +metadata: + name: {{ .Release.Name }}-webhook-ready + namespace: {{ .Release.Namespace }} + annotations: + helm.sh/hook: post-install,post-upgrade + helm.sh/hook-weight: "0" + helm.sh/hook-delete-policy: before-hook-creation,hook-succeeded +roleRef: + apiGroup: rbac.authorization.k8s.io + kind: Role + name: {{ .Release.Name }}-webhook-ready +subjects: + - kind: ServiceAccount + name: {{ .Release.Name }}-webhook-ready + namespace: {{ .Release.Namespace }} +--- +apiVersion: batch/v1 +kind: Job +metadata: + name: {{ .Release.Name }}-webhook-ready + namespace: {{ .Release.Namespace }} + annotations: + helm.sh/hook: post-install,post-upgrade + helm.sh/hook-weight: "10" + helm.sh/hook-delete-policy: before-hook-creation,hook-succeeded +spec: + backoffLimit: {{ .Values.webhookReady.backoffLimit | default 2 }} + activeDeadlineSeconds: {{ $deadline }} + template: + spec: + restartPolicy: Never + serviceAccountName: {{ .Release.Name }}-webhook-ready + securityContext: + runAsNonRoot: true + runAsUser: 65532 + runAsGroup: 65532 + seccompProfile: + type: RuntimeDefault + containers: + - name: wait + image: {{ $image }} + imagePullPolicy: {{ if .Values.webhookReady.image.digest }}IfNotPresent{{ else }}Always{{ end }} + resources: + requests: + cpu: 10m + memory: 32Mi + limits: + cpu: 100m + memory: 64Mi + securityContext: + allowPrivilegeEscalation: false + readOnlyRootFilesystem: true + capabilities: + drop: ["ALL"] + command: + - sh + - -c + - | + set -e + ns={{ .Release.Namespace }} + proxy="/api/v1/namespaces/${ns}/services/{{ $resourceName }}/proxy/readyz" + max_attempts={{ $maxAttempts }} + sleep_seconds={{ $sleepSeconds }} + i=0 + last_err="" + until last_err=$(kubectl get --raw "$proxy" 2>&1 >/dev/null); do + i=$((i + 1)) + if [ $i -gt $max_attempts ]; then + echo "timeout: cnpg webhook did not respond through the apiserver proxy after ${max_attempts} attempts (${sleep_seconds}s each)" + echo "last error: ${last_err}" + exit 1 + fi + if [ $((i % 10)) -eq 1 ]; then + echo "attempt $i/${max_attempts}: ${last_err}" + fi + sleep "$sleep_seconds" + done + echo "cnpg webhook is ready" diff --git a/packages/system/postgres-operator/tests/webhook-ready-hook_test.yaml b/packages/system/postgres-operator/tests/webhook-ready-hook_test.yaml new file mode 100644 index 00000000..3253c663 --- /dev/null +++ b/packages/system/postgres-operator/tests/webhook-ready-hook_test.yaml @@ -0,0 +1,328 @@ +suite: cnpg webhook post-install readiness gate + +templates: + - templates/webhook-ready-hook.yaml + - charts/cloudnative-pg/templates/service.yaml + +release: + name: postgres-operator + namespace: cozy-postgres-operator + +tests: + - it: renders four hook objects (SA + Role + RoleBinding + Job) + template: templates/webhook-ready-hook.yaml + asserts: + - hasDocuments: + count: 4 + + - it: every rendered object carries post-install and post-upgrade hook annotations + template: templates/webhook-ready-hook.yaml + asserts: + - documentIndex: 0 + equal: + path: metadata.annotations["helm.sh/hook"] + value: post-install,post-upgrade + - documentIndex: 1 + equal: + path: metadata.annotations["helm.sh/hook"] + value: post-install,post-upgrade + - documentIndex: 2 + equal: + path: metadata.annotations["helm.sh/hook"] + value: post-install,post-upgrade + - documentIndex: 3 + equal: + path: metadata.annotations["helm.sh/hook"] + value: post-install,post-upgrade + + - it: RBAC is created before the Job (hook-weight ordering) + template: templates/webhook-ready-hook.yaml + asserts: + - documentIndex: 0 + equal: + path: kind + value: ServiceAccount + - documentIndex: 0 + equal: + path: metadata.annotations["helm.sh/hook-weight"] + value: "0" + - documentIndex: 1 + equal: + path: kind + value: Role + - documentIndex: 1 + equal: + path: metadata.annotations["helm.sh/hook-weight"] + value: "0" + - documentIndex: 2 + equal: + path: kind + value: RoleBinding + - documentIndex: 2 + equal: + path: metadata.annotations["helm.sh/hook-weight"] + value: "0" + - documentIndex: 3 + equal: + path: kind + value: Job + - documentIndex: 3 + equal: + path: metadata.annotations["helm.sh/hook-weight"] + value: "10" + + - it: RBAC resourceName matches the exact proxy URL segment the Job probes + template: templates/webhook-ready-hook.yaml + asserts: + - documentIndex: 1 + equal: + path: rules[0].resources[0] + value: services/proxy + - documentIndex: 1 + equal: + path: rules[0].resourceNames[0] + value: https:cnpg-webhook-service:webhook-server + - documentIndex: 3 + matchRegex: + path: spec.template.spec.containers[0].command[2] + pattern: "/api/v1/namespaces/\\$\\{ns\\}/services/https:cnpg-webhook-service:webhook-server/proxy/readyz" + + - it: hardcoded service name in the hook matches the vendored cnpg subchart Service (drift guard for make update) + template: charts/cloudnative-pg/templates/service.yaml + asserts: + - equal: + path: metadata.name + value: cnpg-webhook-service + - equal: + path: spec.ports[0].name + value: webhook-server + + - it: Job calls kubectl get --raw on the proxy path + template: templates/webhook-ready-hook.yaml + asserts: + - documentIndex: 3 + matchRegex: + path: spec.template.spec.containers[0].command[2] + pattern: "kubectl get --raw" + + - it: Job image is digest-pinned when webhookReady.image.digest is set + template: templates/webhook-ready-hook.yaml + set: + webhookReady: + image: + repository: docker.io/clastix/kubectl + tag: v1.32 + digest: sha256:b9ef7d8dbe65bcc81a46c09b8dc7543103055021c4f43287bf59e92a8f4fe05c + asserts: + - documentIndex: 3 + equal: + path: spec.template.spec.containers[0].image + value: docker.io/clastix/kubectl:v1.32@sha256:b9ef7d8dbe65bcc81a46c09b8dc7543103055021c4f43287bf59e92a8f4fe05c + + - it: Job image falls back to tag-only when digest is not configured + template: templates/webhook-ready-hook.yaml + set: + webhookReady: + image: + repository: docker.io/clastix/kubectl + tag: v1.32 + digest: "" + asserts: + - documentIndex: 3 + equal: + path: spec.template.spec.containers[0].image + value: docker.io/clastix/kubectl:v1.32 + + - it: backoffLimit defaults to 2 so transient pod-level failures retry instead of killing the HelmRelease + template: templates/webhook-ready-hook.yaml + asserts: + - documentIndex: 3 + equal: + path: spec.backoffLimit + value: 2 + + - it: backoffLimit is overridable + template: templates/webhook-ready-hook.yaml + set: + webhookReady: + backoffLimit: 5 + asserts: + - documentIndex: 3 + equal: + path: spec.backoffLimit + value: 5 + + - it: chart render fails when webhookReady.image.repository is empty + template: templates/webhook-ready-hook.yaml + set: + webhookReady: + image: + repository: "" + tag: v1.32 + asserts: + - failedTemplate: + errorMessage: "webhookReady.image.repository must be set to the container image providing kubectl for the post-install readiness Job" + + - it: chart render fails when webhookReady.image.tag is empty + template: templates/webhook-ready-hook.yaml + set: + webhookReady: + image: + repository: docker.io/clastix/kubectl + tag: "" + asserts: + - failedTemplate: + errorMessage: "webhookReady.image.tag must be set for the post-install readiness Job" + + - it: retry loop bounds default when blanked so a wiped override still produces a working Job + template: templates/webhook-ready-hook.yaml + set: + webhookReady: + image: + repository: docker.io/clastix/kubectl + tag: v1.32 + maxAttempts: null + sleepSeconds: null + asserts: + - documentIndex: 3 + matchRegex: + path: spec.template.spec.containers[0].command[2] + pattern: "max_attempts=60" + - documentIndex: 3 + matchRegex: + path: spec.template.spec.containers[0].command[2] + pattern: "sleep_seconds=2" + + - it: Job pod runs non-root with seccomp RuntimeDefault for restricted-PSA clusters + template: templates/webhook-ready-hook.yaml + asserts: + - documentIndex: 3 + equal: + path: spec.template.spec.securityContext.runAsNonRoot + value: true + - documentIndex: 3 + equal: + path: spec.template.spec.securityContext.seccompProfile.type + value: RuntimeDefault + + - it: wait container declares resource requests and limits for predictable scheduling + template: templates/webhook-ready-hook.yaml + asserts: + - documentIndex: 3 + equal: + path: spec.template.spec.containers[0].resources.requests.cpu + value: 10m + - documentIndex: 3 + equal: + path: spec.template.spec.containers[0].resources.requests.memory + value: 32Mi + - documentIndex: 3 + equal: + path: spec.template.spec.containers[0].resources.limits.cpu + value: 100m + - documentIndex: 3 + equal: + path: spec.template.spec.containers[0].resources.limits.memory + value: 64Mi + + - it: Job container drops all capabilities and runs read-only rootfs + template: templates/webhook-ready-hook.yaml + asserts: + - documentIndex: 3 + equal: + path: spec.template.spec.containers[0].securityContext.allowPrivilegeEscalation + value: false + - documentIndex: 3 + equal: + path: spec.template.spec.containers[0].securityContext.readOnlyRootFilesystem + value: true + - documentIndex: 3 + equal: + path: spec.template.spec.containers[0].securityContext.capabilities.drop[0] + value: ALL + + - it: imagePullPolicy is IfNotPresent when digest-pinned, Always when tag-only + template: templates/webhook-ready-hook.yaml + set: + webhookReady: + image: + repository: docker.io/clastix/kubectl + tag: v1.32 + digest: sha256:b9ef7d8dbe65bcc81a46c09b8dc7543103055021c4f43287bf59e92a8f4fe05c + asserts: + - documentIndex: 3 + equal: + path: spec.template.spec.containers[0].imagePullPolicy + value: IfNotPresent + + - it: imagePullPolicy is Always when no digest is configured + template: templates/webhook-ready-hook.yaml + set: + webhookReady: + image: + repository: docker.io/clastix/kubectl + tag: v1.32 + digest: "" + asserts: + - documentIndex: 3 + equal: + path: spec.template.spec.containers[0].imagePullPolicy + value: Always + + - it: retry loop captures and surfaces the last kubectl error message on timeout + template: templates/webhook-ready-hook.yaml + asserts: + - documentIndex: 3 + matchRegex: + path: spec.template.spec.containers[0].command[2] + pattern: 'last_err=\$\(kubectl get --raw' + - documentIndex: 3 + matchRegex: + path: spec.template.spec.containers[0].command[2] + pattern: 'last error: \$\{last_err\}' + + - it: retry loop error message stays in sync when maxAttempts is bumped + template: templates/webhook-ready-hook.yaml + set: + webhookReady: + image: + repository: docker.io/clastix/kubectl + tag: v1.32 + maxAttempts: 90 + sleepSeconds: 3 + asserts: + - documentIndex: 3 + matchRegex: + path: spec.template.spec.containers[0].command[2] + pattern: "max_attempts=90" + - documentIndex: 3 + matchRegex: + path: spec.template.spec.containers[0].command[2] + pattern: "sleep_seconds=3" + - documentIndex: 3 + matchRegex: + path: spec.template.spec.containers[0].command[2] + pattern: 'after \$\{max_attempts\} attempts' + + - it: activeDeadlineSeconds scales with retry bounds so an override raise does not silently cut + template: templates/webhook-ready-hook.yaml + set: + webhookReady: + image: + repository: docker.io/clastix/kubectl + tag: v1.32 + maxAttempts: 180 + sleepSeconds: 3 + asserts: + - documentIndex: 3 + equal: + path: spec.activeDeadlineSeconds + value: 600 + + - it: activeDeadlineSeconds defaults include 60s slack over the default retry window + template: templates/webhook-ready-hook.yaml + asserts: + - documentIndex: 3 + equal: + path: spec.activeDeadlineSeconds + value: 180 diff --git a/packages/system/postgres-operator/values.yaml b/packages/system/postgres-operator/values.yaml index cae3dc53..c408220e 100644 --- a/packages/system/postgres-operator/values.yaml +++ b/packages/system/postgres-operator/values.yaml @@ -3,3 +3,30 @@ cloudnative-pg: create: true image: tag: "1.27.3" +# Image used by the post-install webhook-readiness Job (see templates/webhook-ready-hook.yaml). +# Any image with kubectl on PATH works; the Job calls the apiserver service proxy with the +# hook ServiceAccount's token to confirm the mcluster.cnpg.io webhook is reachable end-to-end +# before the HelmRelease reports Ready, closing the "connection refused" bootstrap race. +# +# The tag is digest-pinned so an upstream retag does not change what runs on every install +# and upgrade across the fleet. Refresh by resolving the current manifest-list digest +# (`docker manifest inspect docker.io/clastix/kubectl:v1.32`) and updating `digest` below. +# renovate: datasource=docker depName=docker.io/clastix/kubectl +webhookReady: + image: + repository: docker.io/clastix/kubectl + tag: v1.32 + digest: sha256:b9ef7d8dbe65bcc81a46c09b8dc7543103055021c4f43287bf59e92a8f4fe05c + # Retry loop bounds for the readiness probe. Defaults total ~120s wall clock. + # Both are `default`-coerced in the template so an override that blanks them still + # produces a working Job. + maxAttempts: 60 + sleepSeconds: 2 + # Pod-level retry budget for transient node/registry failures (image pull rate limit, + # OOM, CNI hiccup). The shell loop inside the container only covers reachability retries. + # Pod retries and activeDeadlineSeconds (wall-clock bound on the whole Job across all + # pod retries) are ANDed, so activeDeadlineSeconds is the shorter of the two gates with + # the defaults above: the 60*2+60 = 180s deadline cuts the Job before backoffLimit=2 + # ever matters if pod-level failures eat more than ~60s of the budget. Raise + # maxAttempts/sleepSeconds alongside backoffLimit when tuning for slow image pulls. + backoffLimit: 2 diff --git a/packages/system/postgres-rd/cozyrds/postgres.yaml b/packages/system/postgres-rd/cozyrds/postgres.yaml index c74c5783..72d013f1 100644 --- a/packages/system/postgres-rd/cozyrds/postgres.yaml +++ b/packages/system/postgres-rd/cozyrds/postgres.yaml @@ -8,7 +8,7 @@ spec: singular: postgres plural: postgreses openAPISchema: |- - {"title":"Chart Values","type":"object","properties":{"replicas":{"description":"Number of Postgres replicas.","type":"integer","default":2},"resources":{"description":"Explicit CPU and memory configuration for each PostgreSQL replica. When omitted, the preset defined in `resourcesPreset` is applied.","type":"object","default":{},"properties":{"cpu":{"description":"CPU available to each replica.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true},"memory":{"description":"Memory (RAM) available to each replica.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true}}},"resourcesPreset":{"description":"Default sizing preset used when `resources` is omitted.","type":"string","default":"micro","enum":["nano","micro","small","medium","large","xlarge","2xlarge"]},"size":{"description":"Persistent Volume Claim size available for application data.","default":"10Gi","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true},"storageClass":{"description":"StorageClass used to store the data.","type":"string","default":""},"external":{"description":"Enable external access from outside the cluster.","type":"boolean","default":false},"version":{"description":"PostgreSQL major version to deploy","type":"string","default":"v18","enum":["v18","v17","v16","v15","v14","v13"]},"postgresql":{"description":"PostgreSQL server configuration.","type":"object","default":{},"properties":{"parameters":{"description":"PostgreSQL server parameters.","type":"object","default":{},"properties":{"max_connections":{"description":"Maximum number of concurrent connections to the database server.","type":"integer","default":100}}}}},"quorum":{"description":"Quorum configuration for synchronous replication.","type":"object","default":{},"required":["maxSyncReplicas","minSyncReplicas"],"properties":{"maxSyncReplicas":{"description":"Maximum number of synchronous replicas allowed (must be less than total replicas).","type":"integer","default":0},"minSyncReplicas":{"description":"Minimum number of synchronous replicas required for commit.","type":"integer","default":0}}},"users":{"description":"Users configuration map.","type":"object","default":{},"additionalProperties":{"type":"object","properties":{"password":{"description":"Password for the user.","type":"string"},"replication":{"description":"Whether the user has replication privileges.","type":"boolean"}}}},"databases":{"description":"Databases configuration map.","type":"object","default":{},"additionalProperties":{"type":"object","properties":{"extensions":{"description":"List of enabled PostgreSQL extensions.","type":"array","items":{"type":"string"}},"roles":{"description":"Roles assigned to users.","type":"object","properties":{"admin":{"description":"List of users with admin privileges.","type":"array","items":{"type":"string"}},"readonly":{"description":"List of users with read-only privileges.","type":"array","items":{"type":"string"}}}}}}},"backup":{"description":"Backup configuration.","type":"object","default":{},"required":["enabled"],"properties":{"destinationPath":{"description":"Destination path for backups (e.g. s3://bucket/path/).","type":"string","default":"s3://bucket/path/to/folder/"},"enabled":{"description":"Enable regular backups.","type":"boolean","default":false},"endpointURL":{"description":"S3 endpoint URL for uploads.","type":"string","default":"http://minio-gateway-service:9000"},"retentionPolicy":{"description":"Retention policy (e.g. \"30d\").","type":"string","default":"30d"},"s3AccessKey":{"description":"Access key for S3 authentication.","type":"string","default":""},"s3SecretKey":{"description":"Secret key for S3 authentication.","type":"string","default":""},"schedule":{"description":"Cron schedule for automated backups.","type":"string","default":"0 2 * * * *"}}},"bootstrap":{"description":"Bootstrap configuration.","type":"object","default":{},"required":["enabled","oldName"],"properties":{"enabled":{"description":"Whether to restore from a backup.","type":"boolean","default":false},"oldName":{"description":"Previous cluster name before deletion.","type":"string","default":""},"recoveryTime":{"description":"Timestamp (RFC3339) for point-in-time recovery; empty means latest.","type":"string","default":""}}}}} + {"title":"Chart Values","type":"object","properties":{"replicas":{"description":"Number of Postgres replicas.","type":"integer","default":2},"resources":{"description":"Explicit CPU and memory configuration for each PostgreSQL replica. When omitted, the preset defined in `resourcesPreset` is applied.","type":"object","default":{},"properties":{"cpu":{"description":"CPU available to each replica.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true},"memory":{"description":"Memory (RAM) available to each replica.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true}}},"resourcesPreset":{"description":"Default sizing preset used when `resources` is omitted.","type":"string","default":"micro","enum":["nano","micro","small","medium","large","xlarge","2xlarge"]},"size":{"description":"Persistent Volume Claim size available for application data.","default":"10Gi","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true},"storageClass":{"description":"StorageClass used to store the data.","type":"string","default":""},"external":{"description":"Enable external access from outside the cluster.","type":"boolean","default":false},"version":{"description":"PostgreSQL major version to deploy","type":"string","default":"v18","enum":["v18","v17","v16","v15","v14","v13"]},"postgresql":{"description":"PostgreSQL server configuration.","type":"object","default":{},"properties":{"parameters":{"description":"PostgreSQL server parameters.","type":"object","default":{},"properties":{"max_connections":{"description":"Maximum number of concurrent connections to the database server.","type":"integer","default":100}}}}},"quorum":{"description":"Quorum configuration for synchronous replication.","type":"object","default":{},"required":["maxSyncReplicas","minSyncReplicas"],"properties":{"maxSyncReplicas":{"description":"Maximum number of synchronous replicas allowed (must be less than total replicas).","type":"integer","default":0},"minSyncReplicas":{"description":"Minimum number of synchronous replicas required for commit.","type":"integer","default":0}}},"users":{"description":"Users configuration map.","type":"object","default":{},"additionalProperties":{"type":"object","properties":{"password":{"description":"Password for the user.","type":"string"},"replication":{"description":"Whether the user has replication privileges.","type":"boolean"}}}},"databases":{"description":"Databases configuration map.","type":"object","default":{},"additionalProperties":{"type":"object","properties":{"extensions":{"description":"List of enabled PostgreSQL extensions.","type":"array","items":{"type":"string"}},"roles":{"description":"Roles assigned to users.","type":"object","properties":{"admin":{"description":"List of users with admin privileges.","type":"array","items":{"type":"string"}},"readonly":{"description":"List of users with read-only privileges.","type":"array","items":{"type":"string"}}}}}}},"backup":{"description":"Backup configuration.","type":"object","default":{},"required":["enabled"],"properties":{"destinationPath":{"description":"Destination path for backups (e.g. s3://bucket/path/).","type":"string","default":"s3://bucket/path/to/folder/"},"enabled":{"description":"Enable regular backups.","type":"boolean","default":false},"endpointURL":{"description":"S3 endpoint URL for uploads.","type":"string","default":"http://minio-gateway-service:9000"},"retentionPolicy":{"description":"Retention policy (e.g. \"30d\").","type":"string","default":"30d"},"s3AccessKey":{"description":"Access key for S3 authentication.","type":"string","default":""},"s3SecretKey":{"description":"Secret key for S3 authentication.","type":"string","default":""},"schedule":{"description":"Cron schedule for automated backups.","type":"string","default":"0 2 * * * *"}}},"bootstrap":{"description":"Bootstrap configuration.","type":"object","default":{},"required":["enabled","oldName"],"properties":{"enabled":{"description":"Whether to restore from a backup.","type":"boolean","default":false},"oldName":{"description":"Previous cluster name before deletion.","type":"string","default":""},"recoveryTime":{"description":"Timestamp (RFC3339) for point-in-time recovery; empty means latest.","type":"string","default":""},"serverName":{"description":"Barman server name (S3 path prefix) used by the original cluster when writing backups. Set this only when the original cluster had an explicit barmanObjectStore.serverName that differed from its Kubernetes resource name.","type":"string","default":""}}}}} release: prefix: postgres- labels: @@ -33,7 +33,7 @@ spec: # labelSelector: # helm.toolkit.fluxcd.io/name: "{reqs[0]['metadata','name']}" - keysOrder: [["apiVersion"], ["appVersion"], ["kind"], ["metadata"], ["metadata", "name"], ["spec", "replicas"], ["spec", "resources"], ["spec", "resourcesPreset"], ["spec", "size"], ["spec", "storageClass"], ["spec", "external"], ["spec", "version"], ["spec", "postgresql"], ["spec", "postgresql", "parameters"], ["spec", "postgresql", "parameters", "max_connections"], ["spec", "quorum"], ["spec", "quorum", "minSyncReplicas"], ["spec", "quorum", "maxSyncReplicas"], ["spec", "users"], ["spec", "databases"], ["spec", "backup"], ["spec", "backup", "enabled"], ["spec", "backup", "retentionPolicy"], ["spec", "backup", "destinationPath"], ["spec", "backup", "endpointURL"], ["spec", "backup", "schedule"], ["spec", "backup", "s3AccessKey"], ["spec", "backup", "s3SecretKey"], ["spec", "bootstrap"], ["spec", "bootstrap", "enabled"], ["spec", "bootstrap", "recoveryTime"], ["spec", "bootstrap", "oldName"]] + keysOrder: [["apiVersion"], ["appVersion"], ["kind"], ["metadata"], ["metadata", "name"], ["spec", "replicas"], ["spec", "resources"], ["spec", "resourcesPreset"], ["spec", "size"], ["spec", "storageClass"], ["spec", "external"], ["spec", "version"], ["spec", "postgresql"], ["spec", "postgresql", "parameters"], ["spec", "postgresql", "parameters", "max_connections"], ["spec", "quorum"], ["spec", "quorum", "minSyncReplicas"], ["spec", "quorum", "maxSyncReplicas"], ["spec", "users"], ["spec", "databases"], ["spec", "backup"], ["spec", "backup", "enabled"], ["spec", "backup", "retentionPolicy"], ["spec", "backup", "destinationPath"], ["spec", "backup", "endpointURL"], ["spec", "backup", "schedule"], ["spec", "backup", "s3AccessKey"], ["spec", "backup", "s3SecretKey"], ["spec", "bootstrap"], ["spec", "bootstrap", "enabled"], ["spec", "bootstrap", "recoveryTime"], ["spec", "bootstrap", "oldName"], ["spec", "bootstrap", "serverName"]] secrets: exclude: [] include: diff --git a/packages/system/vm-instance-rd/cozyrds/vm-instance.yaml b/packages/system/vm-instance-rd/cozyrds/vm-instance.yaml index e93fa86b..e56bd9dd 100644 --- a/packages/system/vm-instance-rd/cozyrds/vm-instance.yaml +++ b/packages/system/vm-instance-rd/cozyrds/vm-instance.yaml @@ -8,7 +8,7 @@ spec: singular: vminstance plural: vminstances openAPISchema: |- - {"title":"Chart Values","type":"object","properties":{"external":{"description":"Enable external access from outside the cluster.","type":"boolean","default":false},"externalMethod":{"description":"Method to pass through traffic to the VM.","type":"string","default":"PortList","enum":["PortList","WholeIP"]},"externalPorts":{"description":"Ports to forward from outside the cluster.","type":"array","default":[22],"items":{"type":"integer"}},"runStrategy":{"description":"Requested running state of the VirtualMachineInstance","type":"string","default":"Always","enum":["Always","Halted","Manual","RerunOnFailure","Once"]},"instanceType":{"description":"Virtual Machine instance type.","type":"string","default":"u1.medium"},"instanceProfile":{"description":"Virtual Machine preferences profile.","type":"string","default":"ubuntu","enum":["alpine","centos.7","centos.7.desktop","centos.stream10","centos.stream10.desktop","centos.stream8","centos.stream8.desktop","centos.stream8.dpdk","centos.stream9","centos.stream9.desktop","centos.stream9.dpdk","cirros","fedora","fedora.arm64","opensuse.leap","opensuse.tumbleweed","rhel.10","rhel.10.arm64","rhel.7","rhel.7.desktop","rhel.8","rhel.8.desktop","rhel.8.dpdk","rhel.9","rhel.9.arm64","rhel.9.desktop","rhel.9.dpdk","rhel.9.realtime","sles","ubuntu","windows.10","windows.10.virtio","windows.11","windows.11.virtio","windows.2k16","windows.2k16.virtio","windows.2k19","windows.2k19.virtio","windows.2k22","windows.2k22.virtio","windows.2k25","windows.2k25.virtio",""]},"disks":{"description":"List of disks to attach.","type":"array","default":[],"items":{"type":"object","required":["name"],"properties":{"bus":{"description":"Disk bus type (e.g. \"sata\").","type":"string"},"name":{"description":"Disk name.","type":"string"}}}},"networks":{"description":"Networks to attach the VM to.","type":"array","default":[],"items":{"type":"object","properties":{"name":{"description":"Network attachment name.","type":"string"}}}},"subnets":{"description":"Deprecated: use networks instead.","type":"array","default":[],"items":{"type":"object","properties":{"name":{"description":"Network attachment name.","type":"string"}}}},"gpus":{"description":"List of GPUs to attach (NVIDIA driver requires at least 4 GiB RAM).","type":"array","default":[],"items":{"type":"object","required":["name"],"properties":{"name":{"description":"The name of the GPU resource to attach.","type":"string"}}}},"cpuModel":{"description":"Model specifies the CPU model inside the VMI. List of available models https://github.com/libvirt/libvirt/tree/master/src/cpu_map","type":"string","default":""},"resources":{"description":"Resource configuration for the virtual machine.","type":"object","default":{},"properties":{"cpu":{"description":"Number of CPU cores allocated.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true},"memory":{"description":"Amount of memory allocated.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true},"sockets":{"description":"Number of CPU sockets (vCPU topology).","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true}}},"sshKeys":{"description":"List of SSH public keys for authentication.","type":"array","default":[],"items":{"type":"string"}},"cloudInit":{"description":"Cloud-init user data.","type":"string","default":""},"cloudInitSeed":{"description":"Seed string to generate SMBIOS UUID for the VM.","type":"string","default":""}}} + {"title":"Chart Values","type":"object","properties":{"external":{"description":"Enable external access from outside the cluster.","type":"boolean","default":false},"externalMethod":{"description":"Method to pass through traffic to the VM.","type":"string","default":"PortList","enum":["PortList","WholeIP"]},"externalPorts":{"description":"Ports to forward from outside the cluster.","type":"array","default":[22],"items":{"type":"integer"}},"externalAllowICMP":{"description":"Whether to accept ICMP traffic to the VM in PortList mode (preserves ping and PMTU discovery). No effect in WholeIP mode. Default true so ping behaves as users expect even when port filtering is in effect.","type":"boolean","default":true},"runStrategy":{"description":"Requested running state of the VirtualMachineInstance","type":"string","default":"Always","enum":["Always","Halted","Manual","RerunOnFailure","Once"]},"instanceType":{"description":"Virtual Machine instance type.","type":"string","default":"u1.medium"},"instanceProfile":{"description":"Virtual Machine preferences profile.","type":"string","default":"ubuntu","enum":["alpine","centos.7","centos.7.desktop","centos.stream10","centos.stream10.desktop","centos.stream8","centos.stream8.desktop","centos.stream8.dpdk","centos.stream9","centos.stream9.desktop","centos.stream9.dpdk","cirros","fedora","fedora.arm64","opensuse.leap","opensuse.tumbleweed","rhel.10","rhel.10.arm64","rhel.7","rhel.7.desktop","rhel.8","rhel.8.desktop","rhel.8.dpdk","rhel.9","rhel.9.arm64","rhel.9.desktop","rhel.9.dpdk","rhel.9.realtime","sles","ubuntu","windows.10","windows.10.virtio","windows.11","windows.11.virtio","windows.2k16","windows.2k16.virtio","windows.2k19","windows.2k19.virtio","windows.2k22","windows.2k22.virtio","windows.2k25","windows.2k25.virtio",""]},"disks":{"description":"List of disks to attach.","type":"array","default":[],"items":{"type":"object","required":["name"],"properties":{"bus":{"description":"Disk bus type (e.g. \"sata\").","type":"string"},"name":{"description":"Disk name.","type":"string"}}}},"networks":{"description":"Networks to attach the VM to.","type":"array","default":[],"items":{"type":"object","properties":{"name":{"description":"Network attachment name.","type":"string"}}}},"subnets":{"description":"Deprecated: use networks instead.","type":"array","default":[],"items":{"type":"object","properties":{"name":{"description":"Network attachment name.","type":"string"}}}},"gpus":{"description":"List of GPUs to attach (NVIDIA driver requires at least 4 GiB RAM).","type":"array","default":[],"items":{"type":"object","required":["name"],"properties":{"name":{"description":"The name of the GPU resource to attach.","type":"string"}}}},"cpuModel":{"description":"Model specifies the CPU model inside the VMI. List of available models https://github.com/libvirt/libvirt/tree/master/src/cpu_map","type":"string","default":""},"resources":{"description":"Resource configuration for the virtual machine.","type":"object","default":{},"properties":{"cpu":{"description":"Number of CPU cores allocated.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true},"memory":{"description":"Amount of memory allocated.","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true},"sockets":{"description":"Number of CPU sockets (vCPU topology).","pattern":"^(\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\\+|-)?(([0-9]+(\\.[0-9]*)?)|(\\.[0-9]+))))?$","anyOf":[{"type":"integer"},{"type":"string"}],"x-kubernetes-int-or-string":true}}},"sshKeys":{"description":"List of SSH public keys for authentication.","type":"array","default":[],"items":{"type":"string"}},"cloudInit":{"description":"Cloud-init user data.","type":"string","default":""},"cloudInitSeed":{"description":"Seed string to generate SMBIOS UUID for the VM.","type":"string","default":""}}} release: prefix: vm-instance- labels: @@ -26,7 +26,7 @@ spec: tags: - compute icon: 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 - keysOrder: [["apiVersion"], ["appVersion"], ["kind"], ["metadata"], ["metadata", "name"], ["spec", "external"], ["spec", "externalMethod"], ["spec", "externalPorts"], ["spec", "runStrategy"], ["spec", "instanceType"], ["spec", "instanceProfile"], ["spec", "disks"], ["spec", "networks"], ["spec", "subnets"], ["spec", "gpus"], ["spec", "cpuModel"], ["spec", "resources"], ["spec", "sshKeys"], ["spec", "cloudInit"], ["spec", "cloudInitSeed"]] + keysOrder: [["apiVersion"], ["appVersion"], ["kind"], ["metadata"], ["metadata", "name"], ["spec", "external"], ["spec", "externalMethod"], ["spec", "externalPorts"], ["spec", "externalAllowICMP"], ["spec", "runStrategy"], ["spec", "instanceType"], ["spec", "instanceProfile"], ["spec", "disks"], ["spec", "networks"], ["spec", "subnets"], ["spec", "gpus"], ["spec", "cpuModel"], ["spec", "resources"], ["spec", "sshKeys"], ["spec", "cloudInit"], ["spec", "cloudInitSeed"]] secrets: exclude: [] include: [] diff --git a/pkg/registry/core/tenantnamespace/rest.go b/pkg/registry/core/tenantnamespace/rest.go index f1ed3fab..4ee97eb0 100644 --- a/pkg/registry/core/tenantnamespace/rest.go +++ b/pkg/registry/core/tenantnamespace/rest.go @@ -17,6 +17,8 @@ import ( apierrors "k8s.io/apimachinery/pkg/api/errors" metainternal "k8s.io/apimachinery/pkg/apis/meta/internalversion" metav1 "k8s.io/apimachinery/pkg/apis/meta/v1" + "k8s.io/apimachinery/pkg/fields" + "k8s.io/apimachinery/pkg/labels" "k8s.io/apimachinery/pkg/runtime" "k8s.io/apimachinery/pkg/runtime/schema" "k8s.io/apimachinery/pkg/types" @@ -24,6 +26,7 @@ import ( "k8s.io/apimachinery/pkg/watch" "k8s.io/apiserver/pkg/endpoints/request" "k8s.io/apiserver/pkg/registry/rest" + "k8s.io/klog/v2" "sigs.k8s.io/controller-runtime/pkg/client" corev1alpha1 "github.com/cozystack/cozystack/pkg/apis/core/v1alpha1" @@ -123,8 +126,18 @@ func (r *REST) Get( return nil, apierrors.NewNotFound(r.gvr.GroupResource(), name) } + // Check if user has access to this namespace + hasAccess, err := r.hasAccessToNamespace(ctx, name) + if err != nil { + return nil, err + } + if !hasAccess { + // Return Forbidden to follow standard K8s RBAC behavior + return nil, apierrors.NewForbidden(r.gvr.GroupResource(), name, fmt.Errorf("access denied")) + } + ns := &corev1.Namespace{} - err := r.c.Get(ctx, types.NamespacedName{Namespace: "", Name: name}, ns, &client.GetOptions{Raw: opts}) + err = r.c.Get(ctx, types.NamespacedName{Namespace: "", Name: name}, ns, &client.GetOptions{Raw: opts}) if err != nil { return nil, err } @@ -143,11 +156,33 @@ func (r *REST) Get( // ----------------------------------------------------------------------------- func (r *REST) Watch(ctx context.Context, opts *metainternal.ListOptions) (watch.Interface, error) { + // Extract user identity once for the lifetime of the watch — it does not + // change between events and rebuilding it per event is wasteful. + u, ok := request.UserFrom(ctx) + if !ok { + return nil, apierrors.NewUnauthorized("user missing in context") + } + username := u.GetName() + groups := make(map[string]struct{}) + for _, group := range u.GetGroups() { + groups[group] = struct{}{} + } + nsList := &corev1.NamespaceList{} - nsWatch, err := r.w.Watch(ctx, nsList, &client.ListOptions{Raw: &metav1.ListOptions{ + + // Build upstream watch options with field and label selectors + rawOpts := &metav1.ListOptions{ Watch: true, ResourceVersion: opts.ResourceVersion, - }}) + } + if opts.FieldSelector != nil { + rawOpts.FieldSelector = opts.FieldSelector.String() + } + if opts.LabelSelector != nil { + rawOpts.LabelSelector = opts.LabelSelector.String() + } + + nsWatch, err := r.w.Watch(ctx, nsList, &client.ListOptions{Raw: rawOpts}) if err != nil { return nil, err } @@ -189,6 +224,30 @@ func (r *REST) Watch(ctx context.Context, opts *metainternal.ListOptions) (watch continue } + // Apply defensive filtering for field and label selectors + if opts.FieldSelector != nil { + if !opts.FieldSelector.Matches(fields.Set{"metadata.name": ns.Name}) { + continue + } + } + if opts.LabelSelector != nil { + if !opts.LabelSelector.Matches(labels.Set(ns.Labels)) { + continue + } + } + + // Check if user has access to this namespace using the cached + // identity — avoids re-extracting user/groups on every event. + hasAccess, err := r.hasAccessToNamespaceForUser(ctx, ns.Name, username, groups) + if err != nil { + klog.ErrorS(err, "Failed to check access for namespace in watch", "namespace", ns.Name) + continue + } + if !hasAccess { + // User doesn't have access, skip this event + continue + } + out := &corev1alpha1.TenantNamespace{ TypeMeta: metav1.TypeMeta{ APIVersion: corev1alpha1.SchemeGroupVersion.String(), @@ -309,6 +368,26 @@ func (r *REST) makeList(src *corev1.NamespaceList, allowed []string) *corev1alph return out } +// matchesSubject checks if a RoleBinding subject matches the user's identity. +// It handles Group, User, and ServiceAccount subjects with proper namespace fallback. +func matchesSubject(subj rbacv1.Subject, bindingNamespace, username string, groups map[string]struct{}) bool { + switch subj.Kind { + case "Group": + _, ok := groups[subj.Name] + return ok + case "User": + return subj.Name == username + case "ServiceAccount": + saNamespace := subj.Namespace + if saNamespace == "" { + saNamespace = bindingNamespace + } + return username == fmt.Sprintf("system:serviceaccount:%s:%s", saNamespace, subj.Name) + default: + return false + } +} + func (r *REST) filterAccessible( ctx context.Context, names []string, @@ -347,22 +426,9 @@ func (r *REST) filterAccessible( subjectLoop: for j := range rbs.Items[i].Subjects { subj := rbs.Items[i].Subjects[j] - switch subj.Kind { - case "Group": - if _, ok = groups[subj.Name]; ok { - allowedNameSet[rbs.Items[i].Namespace] = struct{}{} - break subjectLoop - } - case "User": - if subj.Name == u.GetName() { - allowedNameSet[rbs.Items[i].Namespace] = struct{}{} - break subjectLoop - } - case "ServiceAccount": - if u.GetName() == fmt.Sprintf("system:serviceaccount:%s:%s", subj.Namespace, subj.Name) { - allowedNameSet[rbs.Items[i].Namespace] = struct{}{} - break subjectLoop - } + if matchesSubject(subj, rbs.Items[i].Namespace, u.GetName(), groups) { + allowedNameSet[rbs.Items[i].Namespace] = struct{}{} + break subjectLoop } } } @@ -373,6 +439,60 @@ func (r *REST) filterAccessible( return allowed, nil } +// hasAccessToNamespace checks if the user has access to a single namespace. +// This is optimized for Get/Watch operations where we check one namespace at a time. +// It lists RoleBindings only in the target namespace instead of all cluster RoleBindings. +func (r *REST) hasAccessToNamespace( + ctx context.Context, + namespace string, +) (bool, error) { + u, ok := request.UserFrom(ctx) + if !ok { + return false, fmt.Errorf("user missing in context") + } + groups := make(map[string]struct{}) + for _, group := range u.GetGroups() { + groups[group] = struct{}{} + } + return r.hasAccessToNamespaceForUser(ctx, namespace, u.GetName(), groups) +} + +// hasAccessToNamespaceForUser is the inner check that does not re-extract user +// identity from context. Use this in hot paths (e.g. the Watch loop) where the +// caller has already cached the user name and groups. +func (r *REST) hasAccessToNamespaceForUser( + ctx context.Context, + namespace, username string, + groups map[string]struct{}, +) (bool, error) { + // Check privileged groups + if _, ok := groups["system:masters"]; ok { + return true, nil + } + if _, ok := groups["cozystack-cluster-admin"]; ok { + return true, nil + } + + // List RoleBindings only in the target namespace + rbs := &rbacv1.RoleBindingList{} + err := r.c.List(ctx, rbs, client.InNamespace(namespace)) + if err != nil { + return false, fmt.Errorf("failed to list rolebindings in %s: %w", namespace, err) + } + + // Check if user is in any RoleBinding subjects + for i := range rbs.Items { + for j := range rbs.Items[i].Subjects { + subj := rbs.Items[i].Subjects[j] + if matchesSubject(subj, rbs.Items[i].Namespace, username, groups) { + return true, nil + } + } + } + + return false, nil +} + // ----------------------------------------------------------------------------- // Boiler-plate // ----------------------------------------------------------------------------- diff --git a/pkg/registry/core/tenantnamespace/rest_test.go b/pkg/registry/core/tenantnamespace/rest_test.go index 7f2979bc..52d2a075 100644 --- a/pkg/registry/core/tenantnamespace/rest_test.go +++ b/pkg/registry/core/tenantnamespace/rest_test.go @@ -3,10 +3,20 @@ package tenantnamespace import ( + "context" "testing" corev1 "k8s.io/api/core/v1" + rbacv1 "k8s.io/api/rbac/v1" + apierrors "k8s.io/apimachinery/pkg/api/errors" metav1 "k8s.io/apimachinery/pkg/apis/meta/v1" + "k8s.io/apimachinery/pkg/runtime" + "k8s.io/apimachinery/pkg/runtime/schema" + "k8s.io/apiserver/pkg/authentication/user" + "k8s.io/apiserver/pkg/endpoints/request" + "sigs.k8s.io/controller-runtime/pkg/client/fake" + + corev1alpha1 "github.com/cozystack/cozystack/pkg/apis/core/v1alpha1" ) func TestMakeListSortsAlphabetically(t *testing.T) { @@ -38,3 +48,466 @@ func TestMakeListSortsAlphabetically(t *testing.T) { } } } + +// Security tests for IDOR fix + +func TestHasAccessToNamespace_WithUserAccess(t *testing.T) { + scheme := runtime.NewScheme() + _ = corev1.AddToScheme(scheme) + _ = rbacv1.AddToScheme(scheme) + + ns := &corev1.Namespace{ + ObjectMeta: metav1.ObjectMeta{Name: "tenant-test"}, + } + + rb := &rbacv1.RoleBinding{ + ObjectMeta: metav1.ObjectMeta{ + Name: "test-binding", + Namespace: "tenant-test", + }, + Subjects: []rbacv1.Subject{ + {Kind: "User", Name: "test-user", APIGroup: "rbac.authorization.k8s.io"}, + }, + RoleRef: rbacv1.RoleRef{ + APIGroup: "rbac.authorization.k8s.io", + Kind: "Role", + Name: "test-role", + }, + } + + client := fake.NewClientBuilder(). + WithScheme(scheme). + WithObjects(ns, rb). + Build() + + r := &REST{ + c: client, + gvr: schema.GroupVersionResource{Group: "core.cozystack.io", Version: "v1alpha1", Resource: "tenantnamespaces"}, + } + + u := &user.DefaultInfo{ + Name: "test-user", + Groups: []string{"system:authenticated"}, + } + ctx := request.WithUser(context.Background(), u) + + hasAccess, err := r.hasAccessToNamespace(ctx, "tenant-test") + if err != nil { + t.Fatalf("unexpected error: %v", err) + } + if !hasAccess { + t.Error("expected user to have access, but got false") + } +} + +func TestHasAccessToNamespace_WithoutAccess(t *testing.T) { + scheme := runtime.NewScheme() + _ = corev1.AddToScheme(scheme) + _ = rbacv1.AddToScheme(scheme) + + ns := &corev1.Namespace{ + ObjectMeta: metav1.ObjectMeta{Name: "tenant-test"}, + } + + // RoleBinding for different user + rb := &rbacv1.RoleBinding{ + ObjectMeta: metav1.ObjectMeta{ + Name: "test-binding", + Namespace: "tenant-test", + }, + Subjects: []rbacv1.Subject{ + {Kind: "User", Name: "other-user", APIGroup: "rbac.authorization.k8s.io"}, + }, + RoleRef: rbacv1.RoleRef{ + APIGroup: "rbac.authorization.k8s.io", + Kind: "Role", + Name: "test-role", + }, + } + + client := fake.NewClientBuilder(). + WithScheme(scheme). + WithObjects(ns, rb). + Build() + + r := &REST{ + c: client, + gvr: schema.GroupVersionResource{Group: "core.cozystack.io", Version: "v1alpha1", Resource: "tenantnamespaces"}, + } + + u := &user.DefaultInfo{ + Name: "test-user", + Groups: []string{"system:authenticated"}, + } + ctx := request.WithUser(context.Background(), u) + + hasAccess, err := r.hasAccessToNamespace(ctx, "tenant-test") + if err != nil { + t.Fatalf("unexpected error: %v", err) + } + if hasAccess { + t.Error("expected user to NOT have access, but got true") + } +} + +func TestHasAccessToNamespace_WithGroupAccess(t *testing.T) { + scheme := runtime.NewScheme() + _ = corev1.AddToScheme(scheme) + _ = rbacv1.AddToScheme(scheme) + + ns := &corev1.Namespace{ + ObjectMeta: metav1.ObjectMeta{Name: "tenant-test"}, + } + + rb := &rbacv1.RoleBinding{ + ObjectMeta: metav1.ObjectMeta{ + Name: "test-binding", + Namespace: "tenant-test", + }, + Subjects: []rbacv1.Subject{ + {Kind: "Group", Name: "test-group", APIGroup: "rbac.authorization.k8s.io"}, + }, + RoleRef: rbacv1.RoleRef{ + APIGroup: "rbac.authorization.k8s.io", + Kind: "Role", + Name: "test-role", + }, + } + + client := fake.NewClientBuilder(). + WithScheme(scheme). + WithObjects(ns, rb). + Build() + + r := &REST{ + c: client, + gvr: schema.GroupVersionResource{Group: "core.cozystack.io", Version: "v1alpha1", Resource: "tenantnamespaces"}, + } + + u := &user.DefaultInfo{ + Name: "test-user", + Groups: []string{"system:authenticated", "test-group"}, + } + ctx := request.WithUser(context.Background(), u) + + hasAccess, err := r.hasAccessToNamespace(ctx, "tenant-test") + if err != nil { + t.Fatalf("unexpected error: %v", err) + } + if !hasAccess { + t.Error("expected user to have access via group, but got false") + } +} + +func TestHasAccessToNamespace_SystemMasters(t *testing.T) { + scheme := runtime.NewScheme() + _ = corev1.AddToScheme(scheme) + _ = rbacv1.AddToScheme(scheme) + + ns := &corev1.Namespace{ + ObjectMeta: metav1.ObjectMeta{Name: "tenant-test"}, + } + + client := fake.NewClientBuilder(). + WithScheme(scheme). + WithObjects(ns). + Build() + + r := &REST{ + c: client, + gvr: schema.GroupVersionResource{Group: "core.cozystack.io", Version: "v1alpha1", Resource: "tenantnamespaces"}, + } + + u := &user.DefaultInfo{ + Name: "admin", + Groups: []string{"system:masters"}, + } + ctx := request.WithUser(context.Background(), u) + + hasAccess, err := r.hasAccessToNamespace(ctx, "tenant-test") + if err != nil { + t.Fatalf("unexpected error: %v", err) + } + if !hasAccess { + t.Error("expected system:masters to have access, but got false") + } +} + +func TestHasAccessToNamespace_CozyAdminGroup(t *testing.T) { + scheme := runtime.NewScheme() + _ = corev1.AddToScheme(scheme) + _ = rbacv1.AddToScheme(scheme) + + ns := &corev1.Namespace{ + ObjectMeta: metav1.ObjectMeta{Name: "tenant-test"}, + } + + client := fake.NewClientBuilder(). + WithScheme(scheme). + WithObjects(ns). + Build() + + r := &REST{ + c: client, + gvr: schema.GroupVersionResource{Group: "core.cozystack.io", Version: "v1alpha1", Resource: "tenantnamespaces"}, + } + + u := &user.DefaultInfo{ + Name: "cozy-admin", + Groups: []string{"cozystack-cluster-admin"}, + } + ctx := request.WithUser(context.Background(), u) + + hasAccess, err := r.hasAccessToNamespace(ctx, "tenant-test") + if err != nil { + t.Fatalf("unexpected error: %v", err) + } + if !hasAccess { + t.Error("expected cozystack-cluster-admin to have access, but got false") + } +} + +func TestHasAccessToNamespace_ServiceAccount(t *testing.T) { + scheme := runtime.NewScheme() + _ = corev1.AddToScheme(scheme) + _ = rbacv1.AddToScheme(scheme) + + ns := &corev1.Namespace{ + ObjectMeta: metav1.ObjectMeta{Name: "tenant-test"}, + } + + rb := &rbacv1.RoleBinding{ + ObjectMeta: metav1.ObjectMeta{ + Name: "test-binding", + Namespace: "tenant-test", + }, + Subjects: []rbacv1.Subject{ + {Kind: "ServiceAccount", Name: "test-sa", Namespace: "tenant-test"}, + }, + RoleRef: rbacv1.RoleRef{ + APIGroup: "rbac.authorization.k8s.io", + Kind: "Role", + Name: "test-role", + }, + } + + client := fake.NewClientBuilder(). + WithScheme(scheme). + WithObjects(ns, rb). + Build() + + r := &REST{ + c: client, + gvr: schema.GroupVersionResource{Group: "core.cozystack.io", Version: "v1alpha1", Resource: "tenantnamespaces"}, + } + + u := &user.DefaultInfo{ + Name: "system:serviceaccount:tenant-test:test-sa", + Groups: []string{"system:authenticated"}, + } + ctx := request.WithUser(context.Background(), u) + + hasAccess, err := r.hasAccessToNamespace(ctx, "tenant-test") + if err != nil { + t.Fatalf("unexpected error: %v", err) + } + if !hasAccess { + t.Error("expected service account to have access, but got false") + } +} + +func TestHasAccessToNamespace_ServiceAccountEmptyNamespace(t *testing.T) { + scheme := runtime.NewScheme() + _ = corev1.AddToScheme(scheme) + _ = rbacv1.AddToScheme(scheme) + + ns := &corev1.Namespace{ + ObjectMeta: metav1.ObjectMeta{Name: "tenant-test"}, + } + + // ServiceAccount subject with empty namespace should default to RoleBinding namespace + rb := &rbacv1.RoleBinding{ + ObjectMeta: metav1.ObjectMeta{ + Name: "test-binding", + Namespace: "tenant-test", + }, + Subjects: []rbacv1.Subject{ + {Kind: "ServiceAccount", Name: "test-sa", Namespace: ""}, // Empty namespace + }, + RoleRef: rbacv1.RoleRef{ + APIGroup: "rbac.authorization.k8s.io", + Kind: "Role", + Name: "test-role", + }, + } + + client := fake.NewClientBuilder(). + WithScheme(scheme). + WithObjects(ns, rb). + Build() + + r := &REST{ + c: client, + gvr: schema.GroupVersionResource{Group: "core.cozystack.io", Version: "v1alpha1", Resource: "tenantnamespaces"}, + } + + u := &user.DefaultInfo{ + Name: "system:serviceaccount:tenant-test:test-sa", + Groups: []string{"system:authenticated"}, + } + ctx := request.WithUser(context.Background(), u) + + hasAccess, err := r.hasAccessToNamespace(ctx, "tenant-test") + if err != nil { + t.Fatalf("unexpected error: %v", err) + } + if !hasAccess { + t.Error("expected service account with empty namespace to have access, but got false") + } +} + +func TestGet_WithAccess(t *testing.T) { + scheme := runtime.NewScheme() + _ = corev1.AddToScheme(scheme) + _ = rbacv1.AddToScheme(scheme) + + ns := &corev1.Namespace{ + ObjectMeta: metav1.ObjectMeta{Name: "tenant-test"}, + } + + rb := &rbacv1.RoleBinding{ + ObjectMeta: metav1.ObjectMeta{ + Name: "test-binding", + Namespace: "tenant-test", + }, + Subjects: []rbacv1.Subject{ + {Kind: "User", Name: "test-user", APIGroup: "rbac.authorization.k8s.io"}, + }, + RoleRef: rbacv1.RoleRef{ + APIGroup: "rbac.authorization.k8s.io", + Kind: "Role", + Name: "test-role", + }, + } + + client := fake.NewClientBuilder(). + WithScheme(scheme). + WithObjects(ns, rb). + Build() + + r := &REST{ + c: client, + gvr: schema.GroupVersionResource{Group: "core.cozystack.io", Version: "v1alpha1", Resource: "tenantnamespaces"}, + } + + u := &user.DefaultInfo{ + Name: "test-user", + Groups: []string{"system:authenticated"}, + } + ctx := request.WithUser(context.Background(), u) + + obj, err := r.Get(ctx, "tenant-test", &metav1.GetOptions{}) + if err != nil { + t.Fatalf("unexpected error: %v", err) + } + tn, ok := obj.(*corev1alpha1.TenantNamespace) + if !ok { + t.Fatalf("expected *TenantNamespace, got %T", obj) + } + if tn.Name != "tenant-test" { + t.Errorf("expected name %q, got %q", "tenant-test", tn.Name) + } + if tn.Kind != "TenantNamespace" { + t.Errorf("expected Kind=TenantNamespace, got %q", tn.Kind) + } + if tn.APIVersion != corev1alpha1.SchemeGroupVersion.String() { + t.Errorf("expected APIVersion=%q, got %q", corev1alpha1.SchemeGroupVersion.String(), tn.APIVersion) + } +} + +func TestGet_WithoutAccess(t *testing.T) { + scheme := runtime.NewScheme() + _ = corev1.AddToScheme(scheme) + _ = rbacv1.AddToScheme(scheme) + + ns := &corev1.Namespace{ + ObjectMeta: metav1.ObjectMeta{Name: "tenant-test"}, + } + + // RoleBinding for different user + rb := &rbacv1.RoleBinding{ + ObjectMeta: metav1.ObjectMeta{ + Name: "test-binding", + Namespace: "tenant-test", + }, + Subjects: []rbacv1.Subject{ + {Kind: "User", Name: "other-user", APIGroup: "rbac.authorization.k8s.io"}, + }, + RoleRef: rbacv1.RoleRef{ + APIGroup: "rbac.authorization.k8s.io", + Kind: "Role", + Name: "test-role", + }, + } + + client := fake.NewClientBuilder(). + WithScheme(scheme). + WithObjects(ns, rb). + Build() + + r := &REST{ + c: client, + gvr: schema.GroupVersionResource{Group: "core.cozystack.io", Version: "v1alpha1", Resource: "tenantnamespaces"}, + } + + u := &user.DefaultInfo{ + Name: "test-user", + Groups: []string{"system:authenticated"}, + } + ctx := request.WithUser(context.Background(), u) + + obj, err := r.Get(ctx, "tenant-test", &metav1.GetOptions{}) + if err == nil { + t.Fatal("expected error, got nil") + } + if obj != nil { + t.Errorf("expected nil object, got %v", obj) + } + + // Verify it returns Forbidden to follow standard K8s RBAC behavior + if !apierrors.IsForbidden(err) { + t.Errorf("expected Forbidden error, got %v", err) + } +} + +func TestGet_NonTenantNamespace(t *testing.T) { + scheme := runtime.NewScheme() + _ = corev1.AddToScheme(scheme) + _ = rbacv1.AddToScheme(scheme) + + client := fake.NewClientBuilder(). + WithScheme(scheme). + Build() + + r := &REST{ + c: client, + gvr: schema.GroupVersionResource{Group: "core.cozystack.io", Version: "v1alpha1", Resource: "tenantnamespaces"}, + } + + u := &user.DefaultInfo{ + Name: "test-user", + Groups: []string{"system:masters"}, + } + ctx := request.WithUser(context.Background(), u) + + obj, err := r.Get(ctx, "default", &metav1.GetOptions{}) + if err == nil { + t.Fatal("expected error for non-tenant namespace, got nil") + } + if obj != nil { + t.Errorf("expected nil object, got %v", obj) + } + if !apierrors.IsNotFound(err) { + t.Errorf("expected NotFound error, got %v", err) + } +}